Journal articles on the topic 'Modeling, control and learning of soft robots'

Soft robots have been extensively researched due to their flexible, deformable, and adaptive characteristics. However, compared to rigid robots, soft robots have issues in modeling, calibration, and control in that the innate characteristics of the soft materials can cause complex behaviors due to non-linearity and hysteresis. To overcome these limitations, recent studies have applied various approaches based on machine learning. This paper presents existing machine learning techniques in the soft robotic fields and categorizes the implementation of machine learning approaches in different soft robotic applications, which include soft sensors, soft actuators, and applications such as soft wearable robots. An analysis of the trends of different machine learning approaches with respect to different types of soft robot applications is presented; in addition to the current limitations in the research field, followed by a summary of the existing machine learning methods for soft robots.

Conventional model theories are not suitable to control soft-bodied robots as deformable materials present rapidly changing behaviors. Neuromorphic electronics are now entering the field of robotics, demonstrating that a highly integrated device can mimic the fundamental properties of a sensory synaptic system, including learning and proprioception. This research work focuses on the physical implementation of a reservoir computing-based network to actuate a soft-bodied robot. More specifically, modeling the hysteresis of a shape memory alloy (SMA) using echo state networks (ESN) in real-world situations represents a novel approach to enable soft machines with task-learning. In this work, we show that not only does our ESN model enable our SMA-based robot with locomotion, but it also discovers a successful strategy to do so. Compared to standard control modeling, established either by theoretical frameworks or from experimental data, here, we gained knowledge a posteriori, guided by the physical interactions between the trained model and the controlled actuator, interactions from which striking patterns emerged, and informed us about what type of locomotion would work best for our robot.

Compared with traditional underwater vehicles, bio-inspired fish robots have the advantages of high efficiency, high maneuverability, low noise, and minor fluid disturbance. Therefore, they have gained an increasing research interest, which has led to a great deal of remarkable progress theoretically and practically in recent years. In this review, we first highlight our enhanced scientific understanding of bio-inspired propulsion and sensing underwater and then present the research progress and performance characteristics of different bio-inspired robot fish, classified by the propulsion method. Like the natural fish species they imitate, different types of bionic fish have different morphological structures and distinctive hydrodynamic properties. In addition, we select two pioneering directions about soft robotic control and multi-phase robotics. The hybrid dynamic control of soft robotic systems combines the accuracy of model-based control and the efficiency of model-free control, and is considered the proper way to optimize the classical control model with the intersection of multiple machine learning algorithms. Multi-phase robots provide a broader scope of application compared to ordinary bionic robot fish, with the ability of operating in air or on land outside the fluid. By introducing recent progress in related fields, we summarize the advantages and challenges of soft robotic control and multi-phase robotics, guiding the further development of bionic aquatic robots.

Due to the outstanding characteristics of the large structural flexibility and strong dexterity of soft robots, they have attracted great attention. However, the dynamic modeling and precise control of soft robots face huge challenges. Traditional model-based and model-free control methods find it difficult to obtain a balance between complexity and accuracy. In this paper, a dynamic model of a three-chamber continuous pneumatic manipulator is established based on the modal method. Moreover, a hybrid controller integrating model predictive control (MPC) and iterative learning control (ILC) is proposed, which can simultaneously perform model parameter learning and trajectory tracking control. Experimental results show that the proposed control method can optimize the parameters of the dynamic model in real time with less iterations than the traditional model-free method and have good control performance in trajectory tracking experiments. In the future, the proposed dynamic model and the hybrid controller should be verified on a multi-section manipulator.

The cable-driven soft arm is mostly made of soft material; it is difficult to control because of the material characteristics, so the traditional robot arm modeling and control methods cannot be directly applied to the soft robot arm. In this paper, we combine the data-driven modeling method with the reinforcement learning control method to realize the position control task of robotic soft arm, the method of control strategy based on deep Q learning. In order to solve slow convergence and unstable effect in the process of simulation and migration when deep reinforcement learning is applied to the actual robot control task, a control strategy learning method is designed, which is based on the experimental data, to establish a simulation environment for control strategy training, and then applied to the real environment. Finally, it is proved by experiment that the method can effectively complete the control of the soft robot arm, which has better robustness than the traditional method.

Performing daily interaction tasks such as opening doors and pulling drawers in unstructured environments is a challenging problem for robots. The emergence of soft-bodied robots brings a new perspective to solving this problem. In this paper, inspired by humans performing interaction tasks through simple behaviors, we propose a hierarchical control system for soft arms, in which the low-level controller achieves motion control of the arm tip, the high-level controller controls the behaviors of the arm based on the low-level controller, and the top-level planner chooses what behaviors should be taken according to tasks. To realize the motion control of the soft arm in interacting with environments, we propose two control methods. The first is a feedback control method based on a simplified Jacobian model utilizing the motion laws of the soft arm that are not affected by environments during interaction. The second is a control method based on [Formula: see text]-learning, in which we present a novel method to increase training data by setting virtual goals. We implement the hierarchical control system on a platform with the Honeycomb Pneumatic Networks Arm (HPN Arm) and validate the effectiveness of this system on a series of typical daily interaction tasks, which demonstrates this proposed hierarchical control system could render the soft arms to perform interaction tasks as simply as humans, without force sensors or accurate models of the environments. This work provides a new direction for the application of soft-bodied arms and offers a new perspective for the physical interactions between robots and environments.

Modeling of soft robotics systems proves to be an extremely difficult task, due to the large deformation of the soft materials used to make such robots. Reliable and accurate models are necessary for the control task of these soft robots. In this paper, a data-driven approach using machine learning is presented to model the kinematics of Soft Pneumatic Actuators (SPAs). An Echo State Network (ESN) architecture is used to predict the SPA’s tip position in 3 axes. Initially, data from actual 3D printed SPAs is obtained to build a training dataset for the network. Irregular-intervals pressure inputs are used to drive the SPA in different actuation sequences. The network is then iteratively trained and optimized. The demonstrated method is shown to successfully model the complex non-linear behavior of the SPA, using only the control input without any feedback sensory data as additional input to the network. In addition, the ability of the network to estimate the kinematics of SPAs with different orientation angles θ is achieved. The ESN is compared to a Long Short-Term Memory (LSTM) network that is trained on the interpolated experimental data. Both networks are then tested on Finite Element Analysis (FEA) data for other θ angle SPAs not included in the training data. This methodology could offer a general approach to modeling SPAs with varying design parameters.

Bowden-cable-actuated soft exoskeleton robots are known for their light weight and flexibility of power transmission during rehabilitation training or movement assistance for humans. However, friction-induced nonlinearity of the Bowden transmission cable and gearbox backlash pose great challenges forprecise tracking control of the exoskeleton robot. In this paper, we proposed the design of a learning-based repetitive controller which could compensate for the non-linearcable friction and gearbox backlash in an iterative manner. Unlike most of the previous control schemes, the presented controller does not require apriori knowledge or intensive modeling of the friction and backlash inside the exoskeleton transmission system. Instead, it uses the iterative learning control (ILC)to adaptively update the reference trajectory so that theoutput hysteresis caused by friction and backlashis minimized. In particular, a digital phase-lead compensator wasdesigned and integrated with the ILC to address the issue of backlash delay and improve the stability and tracking performance. Experimental results showed an average of seveniterations for the convergence of learningand a 91.1% reduction in the RMS tracking error (~1.37 deg) compared withthe conventional PD control. The proposed controller design offers promising options for the realization of lightweight, wearable exoskeletons with high tracking accuracies.

Rehabilitation is an area of robotics in which human–robot collaboration occurs, requiring adaptation and compliance. Pneumatic artificial muscles (PAM) are soft actuators that have built-in compliance making them usable for rehabilitation robots. Conversely, compliance arises from nonlinear characteristics and generates obstructions in modeling and controlling actions. It is a critical issue limiting the use of PAM. In this work, multi-input single-output (MISO) inverse modeling and inverse dynamics model learning approaches are combined to obtain a novel nonlinear adaptive control scheme for single PAM-actuated 1-DoF rehabilitation devices, for instance, continuous passive motion (CPM) devices. The objective of the proposed system is to bring an alternative solution to the compliant operation of PAM while performing exercise trajectories, to satisfy requirements such as larger range of motion (ROM) and adaptability to external load impedance variations. The control system combines the operation of a nonlinear autoregressive network with exogenous inputs (NARX)-based inverse dynamics estimator used as a global range controller and cascade PIDs for local position and pressure loops. Implementation results demonstrated the efficacy of the introduced method in terms of compliant operation for dynamic external load variations as well as a stable operation in case of impulsive disturbances. To summarize, a simple but efficient method is illustrated to facilitate the common use of PAM.

Accurate kinematic models are essential for effective control of surgical robots. For tendon driven robots, which are common for minimally invasive surgery, the high nonlinearities in the transmission make modelling complex. Machine learning techniques are a preferred approach to tackle this problem. However, surgical environments are rarely structured, due to organs being very soft and deformable, and unpredictable, for instance, because of fluids in the system, wear and break of the tendons that lead to changes of the system’s behaviour. Therefore, the model needs to quickly adapt. In this work, we propose a method to learn the kinematic model of a redundant surgical robot and control it to perform surgical tasks both autonomously and in teleoperation. The approach employs Feedforward Artificial Neural Networks (ANN) for building the kinematic model of the robot offline, and an online adaptive strategy in order to allow the system to conform to the changing environment. To prove the capabilities of the method, a comparison with a simple feedback controller for autonomous tracking is carried out. Simulation results show that the proposed method is capable of achieving very small tracking errors, even when unpredicted changes in the system occur, such as broken joints. The method proved effective also in guaranteeing accurate tracking in teleoperation.

External disturbance poses the primary threat to robot balance in dynamic environments. This paper provides a learning-based control architecture for quadrupedal self-balancing, which is adaptable to multiple unpredictable scenes of external continuous disturbance. Different from conventional methods which construct analytical models which explicitly reason the balancing process, our work utilized reinforcement learning and artificial neural network to avoid incomprehensible mathematical modeling. The control policy is composed of a neural network and a Tanh Gaussian policy, which implicitly establishes the fuzzy mapping from proprioceptive signals to action commands. During the training process, the maximum-entropy method (soft actor-critic algorithm) is employed to endow the policy with powerful exploration and generalization ability. The trained policy is validated in both simulations and realistic experiments with a customized quadruped robot. The results demonstrate that the policy can be easily transferred to the real world without elaborate configurations. Moreover, although this policy is trained in merely one specific vibration condition, it demonstrates robustness under conditions that were never encountered during training.

The Joint Conference of the 2nd International Conference on Soft Computing and Intelligent Systems and the 5th International Symposium on Advanced Intelligent Systems (SCIS & ISIS 2004) held at Keio University in Yokohama, Japan, on September 21-24, 2004, attracted over 300 papers in fields such as mathematics, urban and transport planning, entertainment, intelligent control, learning, image processing, clustering, neural networks applications, evolutionary computation, system modeling, fuzzy measures, and robotics. The Program Committee requested reviewers in SCIS & ISIS 2004 to select papers for a special issue of the Journal of Advanced Computational Intelligence & Intelligent Informatics (JACIII), with 27 papers accepted for publication in a two-part SCIS & ISIS 2004 special – Vol.9, No.2, containing 13 and the second part containing 14. Paper 1 details tap-changer control using neural networks. Papers 2-5 deal with image processing and recognition – Paper 2 proposing a model of saliency-driven scene learning and recognition and applying its model to robotics, paper 3 discussing breast cancer recognition using evolutionary algorithms, paper 4 covering a revised GMDH-typed neural network model applied to medical image recognition, paper 5 presenting how to compensate for missing information in the acquisition of visual information applied to autonomous soccer robot control. Paper 6 details gene expressions networks for 4 fruit fly development stages. Paper 7 proposes an α-constrained particle swarm optimized for solving constrained optimization problem. Paper 8 develops a fuzzy-neuro multilayer perceptron using genetic algorithms for recognizing odor mixtures. Paper 9 discusses how to integrate symbols into neural networks for the fusion of computational and symbolic processing and its effectiveness demonstrated through simulations. Paper 10 proposes an electric dictionary using a set of nodes and links whose usefulness is verified in experiments. Paper 11 presents a multi-agent algorithm for a class scheduling problem, showing its feasibility through computer simulation. Paper 12 proposes inductive temporal formula specification in system verification, reducing memory and time in the task of system verification. Paper 13 applies an agent-based approach to modeling transport using inductive learning by travelers and an evolutionary approach. The last paper analyzes architectural floor plans using a proposed index classifying floor plans from the user's point of view. We thank reviewers for their time and effort in making these special issues available so quickly, and thank the JACIII editorial board, especially Editor-in-Chief Profs. Hirota and Fukuda and Managing Editor Kenta Uchino, for their invaluable aid and advice in putting these special issues together.

It is difficult to introduce highly versatile automation using robots to handling deformable objects such as thread, cloth, wire, long beams, and thin plates in plant production processes, compared to the handling of rigid objects. Office equipment handles deformable objects such as paper and plastic. Problems unique to these objects is caused by speeding up such equipment and demand for upgrading its accuracy. In agriculture and medical care, automatic, intelligent handling of deformable objects such as fruit and animals has long been desired and practical systems sought. Deformable objects whose handling should be versatiley and accurately automated are classified into two groups based on handling: (A) Flexible, mostly thin, fine objects capable of elastic deformation (B) Soft objects easily crushed, such as soft fruits or animals The problem in handling the first group is controlling object deformation of an infinite degree of freedom with a finite number of manipulated variables. In contrast, a significant problem in handling the second group is often how to handle them without exerting excessive stress and how to handle them safely and reliably. The handling of these two groups differ greatly in mechanics and control theory, and this special issue focuses on the first group — flexible objects — mechanical collection and transport studies, control, and software. Recent studies on their handling are classified into four groups for convenience based on handled objects and types of handling task: (a) Control of deformation, internal force, and vibration or path planning of flexible objects (mainly thin plates and beams) using single or multiple manipulators. (b) Task understanding in insertion of elastic into rigid parts and vice versa, and the study of human skills to help robots accomplish these task. (c) Approaches on improved accuracy, intelligent control, and vibration damping in handling and transfer of sheets and strings with low flexural rigidity, represented by paper or wire. (d) Strategies for grasping and unfolding sheets such as cloth whose flexural rigidity is almost nil. For (a), studies are active on deformation control by two robot hands attempting to grasp cloth. 1-3) In the automobile industry, so-called flexible fixtureless assembly systems are advancing in which two robots process or assemble parts in mid-air without a fixed table to reduce lead time and cost. These systems are mostly developed assuming handled parts are rigid. Nguyen et al. work assuming parts such as sheet metal whose deformation must be taken into consideration.1) Nakagaki et al. propose form estimation that considers even plastic deformation in wire handling by robots, in connection with the development of robots for electric wire installation.4) Many studies cover flexible wire as elastic beams,3-9) but comparatively few focus on bending deformation of thin plates. This special edition includes a paper by Kosuge et al. on thin-plate deformation control. Vibration control of grasped objects becomes important as speed increases. Matsuno kindly contributed his paper on optimum path planning in elastic plate handling. In controlling the deformation of elastic bodies, the mechanics of objects handled is often unknown. This special issue features a paper by Kojima et al. on an approach to this problem by adaptive feed-forward control. For (b), we consider three cases: (1) A cylindrical rigid body inserted into a hole on an elastic plate. (2) An elastic bar inserted into a hole on a rigid body. (3) A tubular elastic body put on a cylindrical rigid body. This special issue carries papers on these problems by Brata et al., Matsuno et al., and Hirai. For (2), a paper by Nakagaki et al.10) covers electric wire installation. For (3), the paper by Shima et al.11) covers insertion of a rigid axis into an elastic hose. Robot skill acquisition is an important issue in robotics in general, and the above papers should prove highly interesting and information because they treat studies by comparing robot and human skills in accomplishing work and acquiring concrete skills knowledge. For (c), attempts are made to theoretically analyze sheet handling mechanisms and control developed based on trial and error, and to structure design theory based on such analysis. These attempts are related to the increased accuracy and speed and enhanced intelligence of sheet-handling office automation equipment such as printers, facsimile machines, copiers, and automated teller machines. Yoshida et al. conducted a series of studies on the effects of guides forming paper feed paths and of inertia force of paper by approximating sheets with a chain of discrete masses and springs.12-14) This special edition also features a study on sheet sticking and jamming. Okuna et al. handles a system of similar nature, mechanical studying the form of paper guides.15) Introducing mechanisms to control the positioning of sheets is effective in raising sheet transfer accuracy. Feedback control that regulates feed roller skew angle as a manipulated variable is proposed.16) Increased reliability in separating single sheets from stacked effectively reduces the malfunction rate in sheet-handling equipment. Ways of optimizing the form of sheet-separation rollers17) and estimating frictional force between separation gates and sheets 18) are also proposed. This special issue contains a proposal by Nakazawa et al. of a mechanism that uses reactive sheet buckling force, made in connection with development of a newspaper page turner for the disabled as technology for separating single sheets. Dry frictional force is most widely used for transporting sheets, but is not stable and may even act as an obstacle to improving accuracy. Niino et al. propose a sheet transfer mechanism that uses electrostatic force.19) For improving the accuracy of flexible wire transmission, this special issue carries a study on transporting flexible thin wire through tension control at multiple points, from a study by Morimitsu et al. on optical fiber installation. The thickness of wire used in equipment is becoming increasingly slim and flexible, along with the equipment it is used in. Tension control in the production process is an important factor in the manufacture of such thin wire. Production efficiency constantly calls for increased transfer speed. It has thus become important to estimate air resistance and inertia and to measure and control the tension of running wire. Studies20,21) by Batra, Fraser, et al. which deal the motion of string in the spinning process provide good examples for learning analytical techniques for air drag and inertia. In string vibration where inertia dominates, attempts are made to control vibration by boundary shaking22,23) and feed-forward/back control.24) For (d), highly versatile robots for handling cloth are being developed, and the software technology for automatic cloth selection and unfolding by robot hands is a popular topic.25-27) Ono et al. comment on the nature of problems in developing intelligent systems for handling cloth and similar objects whose bending rigidity is low and which readily fold and overlap—a paper that will prove a good reference in basic approaches in this field. Mechanical analyses are indispensable to studies on (a) through (c). In contrast, information technology such as characteristic variable measurement, image processing, and discrimination, rather than mechanical analyses, play an important roles in studies on (d). This special issue features a study by Hamashima, Uraya et al. on cloth unfolding as an example of such studies. Studies up to now largely assumed that properties of grasped objects did not change environmental influences such as temperature and humidity. Such influence is often, however, a major factor in handling fiber thread and cloth. This special issue has a paper contributed by Taylor, who studies handling method to prevent influence by such environmental factors. The objective of this special issue will have been achieved if it aids those studying the handling of flexible objects by providing approaches and methodologies of researchers whose target objects differ and if it aids those planning to take up study in this field by providing a general view of this field. References: 1) Nguyen, W. and Mills, J., ""Multi-Robot Control For Plexible Fixtureless Assembly of Flexible Sheet Metal Auto Body Parts,"" Proceedings of the 1996 IEEE International Conference on Robotics and Automation, 2340-2345, (1996). 2) Sun, D. and Shi, X. and Liu, Y., ""Modeling and Cooperation of Two-Arm Robotic System Manipulating a Deformable Object,"" Proceedings of the 1996 IEEE International Conference on Robotics and Automation, 2346-2351, (1996). 3) Kosuge, K., Sakaki, M., Kanitani, K., Yoshida, H. and Fukuda, T., ""Manipulation of a Flexible Object by Dual Manipulators,"" IEEE International Conference on Robotics and Automation, 318-323, (1995). 4) Nakagaki, H., Kitagaki, K., Ogasawara, T. and Tukune H., ""Handling of a Flexible Wire -Detecting a Deformed Shape of the Wire by Vision and a Force Sensor,"" Annual Conference on Robotics and Mechatronics (ROBOMEC'96), 207-210, (1996). 5) Wakamatsu, H., Hirai, S. and Iwata, K., ""Static Analysis of Deformable Object Grasping Based on Bounded Force Closure,"" Trans. of JSML, 84-618 (C), 508-515, (1998). 6) Katoh, R. and Fujmoto, T., ""Study on Deformation of Elastic Object By Manipulator -Path Planning of End -Effector-,"" J. of the Robotics Society of Japan, 13-1, 157-160, (1995). 7) Yukawa, T., Uohiyama, M. and Inooka, M., ""Stability of Control System in Handling a Flexible Object by Rigid Arm Robots,"" JSME Annual Conference on Robotics and Mechatronics (ROBOMEC'95), 169-172, (1995). 8) Yukawa, T., Uohiyama, M. and Cbinata, G., ""Handling of a Vibrating Flexible Structure by a Robot,"" Trans. JSME, 61-583, 938-943, (1995). 9) Sun, D. and Liu, Y., ""Modeling and Impedance Control of a Two-Manipulator System Handling a Flexible Beam,"" Trans. of the ASME, 119, 736-742, (1997). 10) Nakagaki, H., Kitagaki, K. and Tukune, H., ""Contact Motion in Inserting a Flexible Wire into a Hole,"" Annual Conference on Robotics and Mechatronics (ROBOMEC'95), 175-178, (1995). 11) Shimaji, S., Brata, A. and Hattori, H., ""Robot Skill in Assembling a Cylinder into an Elastic Hose,"" Annual Conference on Robotics and Mechatronics (ROBOMEC'95), 752-755, (1995). 12) Yoshida, K. and Kawauchi, M., ""The Analysis of Deformation and Behavior of Flexible Materials (1st Reprt, Study of Spring-Mass Beam Model of the Sheet,"" Trans. of JSME, 58-552, 1474-1480, (1992). 13) Yoshida, K., ""Analysis of Deformation and Behavior of Flexible Materials (2nd Report, Static Analysis for Deformation of the Sheet in the Space Formed by Guide Plates),"" Trans. JSME, 60-570, 501-507, (1994). 14) Yoshida, K., ""Dynamic Analysis of Sheet Defofmation Using Spring-Mass-Beam Model,"" Trans. JSME, 63-615, 3926-3932 (1997). 15) Okuna, K., Nishigaito, T. and Shina, Y., ""Analysis of Paper Deformation Considering Guide Friction (Improvement of Paper Path for Paper-Feeding Mechanism),"" Trans. JSME, 60-575, 2279-2284, (1994). 16) Fujimura, H. and Ono, K., ""Analysis of Paper Motion Driven by Skew-Roll Paper Feeding System,"" Trans. JSME, 62-596, 1354-1360, (1996). 17) Shima, Y., Hattori, S., Kobayashi, Y. and Ukai, M., ""Optimum of Gate-Roller Shape in Paper Isolating Methods,"" Conference of Information, Intelligence and Precision Equipment (IIP'96), 61-62, (1996). 18) Suzuki, Y, Hattori, S., Shima, Y. and Ukai, M., ""Contact Analysis of Paper in Gate-Roller Handling Method"", Conference on Information, Intelligence and Precision Equipment (IIP'95), 19-20, (1995). 19) Niino, T., Egawa, S. and Higuchi, T., ""An Electrostatic Paper Feeder,"" J. of the Japan Society for Precision Engineering, 60-12,1761-1765, (1994). 20) Batra, S., Ghosh, T. and Zeidman, M., ""An Integrated Approach to Dynamic Analysis of the Ring Spinning Process , PartII: With Air Drag,"" Textile Research Journal, 59, 416-424, (1989). 21) Fraser, W., Ghosh, T. and Batra, S., ""On Unwinding Yarn from a Cylindrical Package,"" Proceedings of Royal Society of London, A, 436, 479-438, (1992). 22) Jacob, S., ""Control of Vibrating String Using Impedance Matching,"" Proceedings of the American Control Conference (San Francisco),468-472, (1993). 23) Lee, S. and Mote, C., ""Vibration Control of an Axially Moving String by Boundary Control,"" Trans. of the ASME, J. of Dynamic Systems, Measurement, and Control, 118, 66-74, (1996). 24) Ying, S. and Tan, C., ""Active Vibration Control of the Axially Moving String Using Space Feedforward and Feedback Controllers,"" Trans. ASME, J. of Vibration and Acoustics, 118, 306-312, (1996). 25) Ono, E., Ichijo, H. and Aisaka, N., ""Flexible Robotic Hand for Handling Fabric Pieces in Garment Manufacture,"" International Journal of Clothing Science and Technology, 4-5,18-23, (1992). 26) Paraschidis, K., Fahantidis, N, Petridis, V., Doulgeri, Z., Petrou, L. and Hasapis, G, ""A Robotic System for Handling Textile and Non Rigid Flat Materials,"" Computers in Industry, 26, 303-313, (1995). 27) Fahantidis, N., Paraschidis, K, Petridis, V., Doulgeri, Z., Petrou, L. and Hasapis, G., ""Robot Handling of Flat Textile Materials,"" IEEE Robotics & Automation Magazine, 4-1, 34-41, (1997).

In engineering practice we often have to deal with complex systems, where the conventional approaches for understanding and predicting the behavior of the system can prove to be inadequate. Hence, the researchers try to put some intelligence into the system. The term intelligence in this context still more or less remains a mysterious phenomenon and can be characterized by different abilities of the system or machine, such as adaptation, decision-making, learning, recognition, diagnostics, autonomy, etc. Many of the new results related to this area are published in Journals and in International Conference Proceedings. One such conference is the "IEEE International Conference on Intelligent Engineering Systems". The fourth conference in this series (INES 2000) took place in Portoroz, Slovenia, on September 17-19,2000. There were around eighty participants from eighteen countries around the world. We are glad that so many authors have contributed to ideas related to the issues at the conference. Many of the papers were about applications and design, and others on more theoretical aspects of intelligent systems. This variety made the selection of papers for this special issue very difficult. Eight papers have been selected in the end, which cover different aspects of intelligent engineering systems. It should be pointed out that the respective authors were also kind to revise and update the presented papers for this special issue. The first paper deals with the manipulation problem where the motion changes depending on the state of the system as it is the case in the finger gaiting applications. To solve it the semi-stratified control theory using smooth motion planning is used. The proposed concept combines the stratified motion planning with the unconstrained finger allocations. In the second paper a special branch of Soft Computing developed for the control of mechanical devices is described. It reduces the number of free parameters and computational complexity. For illustration of the efficiency of the proposed adaptive control, a simulation of polishing with a 3 DOF robot is given. The next paper discusses the force control of redundant robots in an unstructured environment. A special attention is given to the decoupling of the task space and null space motion. For that the minimal null space approach is used. The proposed impedance controller assures good task space performances and minimizes the disturbances caused by obstacles. The performance of the proposed controllers has been evaluated by the simulation and by experiments on a real robot. The forth paper presents some advanced modeling approaches and methods. As one of the key issues a manufacturing process model fully associative with form feature based part model has been introduced. The motivation has been that the low level integration of design and manufacturing of mechanical parts, as identified by the authors, is still a main drawback of efficient application of expensive modeling systems. The proposed method allows for creating part model simultaneously with their analysis of machineability. The next paper discusses the design of fractal-order discrete-time controllers. Some approaches to implement fractal derivatives and integrals are analyzed. As the application of the theory of fractional calculus is rather new, many aspects remain to be investigated. The sixth paper demonstrates how to map classical dictionaries and similar structured data to a hypertext structure that is more suitable for the modern media. To achieve the new shape automatically, the HiLog language is used. The automated mapping is illustrated by an example based on Oxford Dictionary of Modern English. In the seventh paper a humanoid robotics shoulder is compared to the human shoulder. First, the capabilities of the robotics shoulder are analyzed and next, using the optical measurement system the human shoulder movements have been measured and analyzed. The last paper discusses the bias-variance tests on multi-layer perception. The performance of Bayesian neural networks is compared with the performance of neural networks trained with a gradient method. Additionally, it is analyzed if it is possible to use a number of networks in committee trained with gradient descent to achieve the performance of a Bayesian network.

Building intelligent systems has been one of the great challenges since the early days of human culture. From the second half of the 18th century, two revolutionary changes played the key role in technical development, hence in creating engineering and intelligent engineering systems. The industrial revolution was made possible through technical advances, and muscle power was replaced by machine power. The information revolution of our time, in turn, canbe characterized as the replacement of brain power by machine intelligence. The technique used to build engineering systems and replace muscle power can be termed "Hard Automation"1) and deals with industrial processes that are fixed and repetitive in nature. In hard automation, the system configuration and the operations are fixed and cannot be changed without considerable down-time and cost. It can be used, however, particularly in applications calling for fast, accurate operation, when manufacturing large batches of the same product. The "intelligent" area of automation is "Soft Automation," which involves the flexible, intelligent operation of an automated process. In flexible automation, the task is programmable and a work cell must be reconfigured quickly to accommodate a product change. It is particularly suitable for plant environments in which a variety of products is manufactured in small batches. Processes in flexible automation may have unexpected or previously unknown conditions, and would require a certain degree of "machine" intelligence to handle them.The term machine intelligence has been changing with time and is machinespecific, so intelligence in this context still remains more or less a mysterious phenomenon. Following Prof. Lotfi A. Zadeh,2) we consider a system intelligent if it has a high machine intelligence quotient (MIQ). As Prof. Zadeh stated, "MIQ is a measure of intelligence of man-made systems," and can be characterized by its well defined dimensions, such as planning, decision making, problem solving, learning reasoning, natural language understanding, speech recognition, handwriting recognition, pattern recognition, diagnostics, and execution of high level instructions.Engineering practice often involves complex systems having multiple variable and multiple parameter models, sometimes with nonlinear coupling. The conventional approaches for understanding and predicting the behavior of such systems based on analytical techniques can prove to be inadequate, even at the initial stages of setting up an appropriate mathematical model. The computational environment used in such an analytical approach is sometimes too categoric and inflexible in order to cope with the intricacy and complexity of real-world industrial systems. It turns out that, in dealing with such systems, one must face a high degree of uncertainty and tolerate great imprecision. Trying to increase precision can be very costly.In the face of the difficulties above, Prof. Zadeh proposes a different approach for Machine Intelligence. He separates Hard Computing techniques based Artificial Intelligence from Soft Computing techniques based Computational Intelligence.•Hard computing is oriented toward the analysis and design of physical processes and systems, and is characterized by precision, formality, and categorization. It is based on binary logic, crisp systems, numerical analysis, probability theory, differential equations, functional analysis, mathematical programming approximation theory, and crisp software.•Soft computing is oriented toward the analysis and design of intelligent systems. It is based on fuzzy logic, artificial neural networks, and probabilistic reasoning, including genetic algorithms, chaos theory, and parts of machine learning, and is characterized by approximation and dispositionality.In hard computing, imprecision and uncertainty are undesirable properties. In soft computing, the tolerance for imprecision and uncertainty is exploited to achieve an acceptable solution at low cost, tractability, and a high MIQ. Prof. Zadeh argues that soft rather than hard computing should be viewed as the foundation of real machine intelligence. A center has been established - the Berkeley Initiative for Soft Computing (BISC) - and he directs it at the University of California, Berkeley. BISC devotes its activities to this concept.3) Soft computing, as he explains2),•is a consortium of methodologies providing a foundation for the conception and design of intelligent systems,•is aimed at formalizing of the remarkable human ability to make rational decision in an uncertain, imprecise environment.The guiding principle of soft computing, given by Prof. Zadeh2) is: Exploit the tolerance for imprecision, uncertainty, and partial truth to achieve tractability, robustness, low solution cost, and better rapport with reality.Fuzzy logic is mainly concerned with imprecision and approximate reasoning, neurocomputing mainly with learning and curve fitting, genetic computation mainly with searching and optimization and probabilistic reasoning mainly with uncertainty and propagation of belief. The constituents of soft computing are complementary rather than competitive. Experience gained over the past decade indicates that it can be more effective to use them combined, rather than exclusively.Based on this approach, machine intelligence, including artificial intelligence and computational intelligence (soft computing techniques) is one pillar of Intelligent Engineering Systems. Hundreds of new results in this area are published in journals and international conference proceedings. One such conference, organized in Budapest, Hungary, on September 15-17, 1997, was titled'IEEE International Conference on Intelligent Engineering Systems 1997' (INES'97), sponsored by the IEEE Industrial Electronics Society, IEEE Hungary Section, Bá{a}nki Doná{a}t Polytechnic, Hungary, National Committee for Technological Development, Hungary, and in technical cooperation with the IEEE Robotics & Automation Society. It had around 100 participants from 29 countries. This special issue features papers selected from those papers presented during the conference. It should be pointed out that these papers are revised and expanded versions of those presented.The first paper discusses an intelligent control system of an automated guided vehicle used in container terminals. Container terminals, as the center of cargo transportation, play a key role in everyday cargo handling. Learning control has been applied to maintaining the vehicle's course and enabling it to stop at a designatedlocation. Speed control uses conventional control. System performance system was evaluated by simulation, and performance tests slated for a test vehicle.The second paper presents a real-time camera-based system designed for gaze tracking focused on human-computer communication. The objective was to equip computer systems with a tool that provides visual information about the user. The system detects the user's presence, then locates and tracks the face, nose and both eyes. Detection is enabled by combining image processing techniques and pattern recognition.The third paper discusses the application of soft computing techniques to solve modeling and control problems in system engineering. After the design of classical PID and fuzzy PID controllers for nonlinear systems with an approximately known dynamic model, the neural control of a SCARA robot is considered. Fuzzy control is discussed for a special class of MIMO nonlinear systems and the method of Wang generalized for such systems.The next paper describes fuzzy and neural network algorithms for word frequency prediction in document filtering. The two techniques presented are compared and an alternative neural network algoritm discussed.The fifth paper highlights the theory of common-sense knowledge in representation and reasoning. A connectionist model is proposed for common-sense knowledge representation and reasoning, and experimental results using this method presented.The next paper introduces an expert consulting system that employs software agents to manage distributed knowledge sources. These individual software agents solve users' problems either by themselves or thorough mutual cooperation.The last paper presents a methodology for creating and applying a generic manufacturing process model for mechanical parts. Based on the product model and other up-to-date approaches, the proposed model involves all possible manufacturing process variants for a cluster of manufacturing tasks. The application involves a four-level model structure and Petri net representation of manufacturing process entities. Creation and evaluation of model entities and representation of the knowledge built in the shape and manufacturing process models are emphasised. The proposed process model is applied in manufacturing process planning and production scheduling.References:1) C. W. De Silva, "Automation Intelligence," Engineering Application of Artificial Intelligence, 7-5, 471-477, (1994).2) L. A. Zadeh, "Fuzzy Logic, Neural Networks and Soft Computing," NATO Advanced Studies Institute on Soft Computing and Its Application, Antalya, Turkey, (1996).3) L. A. Zadeh, "Berkeley Initiative_in Soft Computing," IEEE Industrial Electronics Society Newsletter. 41-3, 8-10, (1994).

Abstract The HASEL actuator is a cutting-edge soft robot compound that is well suited for tasks in unstructured, dynamic environments and has the penitential for superiorly comfortable and smooth human-robot Interaction. However, the nonlinear relation between the input voltage, output strain of the actuators, and the difficulty of analytical modelling makes it hard to design its control software due to the various source of kinematic noises. Machine learning technics, however, which are invented to study the implicit relations in multiparameter problems that do not require pre-existing knowledge, are well suited for HASEL actuators. Traditionally, researchers consider the behavior of this time-dependent system as a sequence of consecutive statuses and use machine learning to enhance conventional algorithms that consume previous and current status and target and adjust the system using varying control input. However, HASEL actuators’ unique propriety of self-stable and negligible lag in response to input changing makes it possible to consider the spatial path of the structure as a whole and control it based on pattern matching. Introducing Recurrent Neural Networks (RNN) and multilayer perceptron (MLP), this paper presents a pattern-matching-based predictive control algorithm for the HASEL actuator system with acceptable size and high accuracy.

Soft robots offer many advantages over traditional rigid robots. However, soft robots can be difficult to control with standard control methods. Fortunately, evolutionary algorithms can offer an elegant solution to this problem. Instead of creating controls to handle the intricate dynamics of these robots, we can simply evolve the controls using a simulation to provide an evaluation function. In this article, we show how such a control paradigm can be applied to an emerging field within soft robotics: robots based on tensegrity structures. We take the model of the Spherical Underactuated Planetary Exploration Robot ball (SUPERball), an icosahedron tensegrity robot under production at NASA Ames Research Center, develop a rolling locomotion algorithm, and study the learned behavior using an accurate model of the SUPERball simulated in the NASA Tensegrity Robotics Toolkit. We first present the historical-average fitness-shaping algorithm for coevolutionary algorithms to speed up learning while favoring robustness over optimality. Second, we use a distributed control approach by coevolving open-loop control signals for each controller. Being simple and distributed, open-loop controllers can be readily implemented on SUPERball hardware without the need for sensor information or precise coordination. We analyze signals of different complexities and frequencies. Among the learned policies, we take one of the best and use it to analyze different aspects of the rolling gait, such as lengths, tensions, and energy consumption. We also discuss the correlation between the signals controlling different parts of the tensegrity robot.

Controlling soft robots is a significant challenge due to the nonlinear elastic nature of the soft materials that conform their structure. This paper studies the identification and control problems of a novel two-degrees-of-freedom, tendon-actuated, soft robotic arm. A decoupled identification approach is presented; later, a fractional order control strategy is proposed and tested experimentally, in comparison with PI solutions. The simulation and experimental results show the goodness of the modeling and control approaches discussed.

With the rise in new soft robotic applications, the control requirements increase. Therefore, precise control methods for soft robots are required. However, the dynamic control of soft robots, which is required for fast movements, is still an open topic and will be discussed here. In this contribution, one kinematic and two dynamic control methods for soft robots are examined. Thereby, an LQI controller with gain scheduling, which is new to soft robotic applications, and an MPC controller are presented. The controllers are compared in a simulation regarding their accuracy and robustness. Additionally, the required implementation effort and computational effort is examined. For this purpose, the trajectory tracking control of a simple soft robot is studied for different trajectories. The soft robot is beam-shaped and tendon-actuated. It is modeled using the piecewise constant curvature model, which is one of the most popular modeling techniques in soft robotics. In this paper, it is shown that all three controllers are able to follow the examined trajectories. However, the dynamic controllers show much higher accuracy and robustness than the kinematic controller. Nevertheless, it should be noted that the implementation and computational effort for the dynamic controllers is significantly higher. Therefore, kinematic controllers should be used if movements are slow and small oscillations can be accepted, while dynamic controllers should be used for faster movements with higher accuracy or robustness requirements.

Rehabilitation and nursing-care robots have become one of the prevalent methods for assistant treatment of motor disorder patients in the field of medical rehabilitation. Traditional rehabilitation robots are mostly made of rigid materials, which significantly limits their application for medical rehabilitation and nursing-care. Soft robots show great potential in the field of rehabilitation robots because of their inherent compliance and safety when they interact with humans. In this paper, we conduct a systematic summary and discussion on the soft rehabilitation and nursing-care robots. This study reviews typical mechanical structures, modeling methods, and control strategies of soft rehabilitation and nursing-care robots in recent years. We classify soft rehabilitation and nursing-care robots into two categories according to their actuation technology, one is based on tendon-driven actuation and the other is based on soft intelligent material actuation. Finally, we analyze and discuss the future directions and work about soft rehabilitation and nursing-care robots, which can provide useful guidance and help on the development of advanced soft rehabilitation and nursing-care robots.

In this review paper, we are interested in the models and algorithms that allow generic simulation and control of a soft robot. First, we start with a quick overview of modeling approaches for soft robots and available methods for calculating the mechanical compliance, and in particular numerical methods, like real-time Finite Element Method (FEM). We also show how these models can be updated based on sensor data. Then, we are interested in the problem of inverse kinematics, under constraints, with generic solutions without assumption on the robot shape, the type, the placement or the redundancy of the actuators, the material behavior… We are also interested by the use of these models and algorithms in case of contact with the environment. Moreover, we refer to dynamic control algorithms based on mechanical models, allowing for robust control of the positioning of the robot. For each of these aspects, this paper gives a quick overview of the existing methods and a focus on the use of FEM. Finally, we discuss the implementation and our contribution in the field for an open soft robotics research.

Abstract Underwater soft robotics is receiving growing popularity within the scientific community, thanks to its prospective capability of tackling challenges that are hard to deal with using traditional rigid technologies, especially while interacting with an unstructured environment. Recently, we proposed a multi-module underwater robotic system with deformable propellers, inspired by bacteria morphology [1]. Here, the same bio-inspired modular structure is employed to perform manipulation tasks, in order to design a multi-functional integrated system. Employing the Geometric Variable Strain Approach, we simulate a scenario where the flagellated robot moves towards a preferred target and, using the same soft appendages, it hooks to it, simulating a monitoring task. The modeling approach and the design allow the Embodied Intelligence principles to exploit the robot’s surrounding environment (water), the shape of the grip-target and the robot’s compliant nature to mediate effective navigation and safe interaction with the target, using few control inputs.

The soft robot is a kind of continuum robot, which is mainly made of soft elastic material or malleable material. It can be continuously deformed in a limited space, and can obtain energy in large bending or high curvature distortion. It has obvious advantages such as high security of human-computer interaction, strong adaptability of unstructured environment, high driving efficiency, low maintenance cost, etc. It has wide application prospects in the fields of industrial production, defense military, medical rehabilitation, exploration, and so on. From the perspective of the bionic mechanism, this paper introduces the soft robots corresponding to insect crawling, snake crawling, fish swimming, elephant trunk, arm, etc. According to different driving modes, the soft robots can be classified into pneumatic-hydraulic driven, intelligent material driven, chemical reaction driven, and so on. The mechanical modeling, control strategy, material, and manufacturing methods of soft robot are summarized, and the application fields of soft robot are introduced. This paper analyzes the main challenges faced by the research on the key technologies of soft robots, summarizes and analyzes them, and puts forward the prospects for the future research of soft robots. The development trend of the future is to develop the soft robot with the characteristics of micro-scale, rigid-flexible coupling, variable stiffness, multi-functional, high integration, and intelligence of driving sensor control.

Future application areas for humanoid robots range from the household, to agriculture, to the military, and to the exploration of space. Service applications such as these must address a changing, unstructured environment, a collaboration with human clients, and the integration of manual dexterity and mobility. Control frameworks for service-oriented humanoid robots must, therefore, accommodate many independently challenging issues including: techniques for configuring networks of sensorimotor resources; modeling tasks and constructing behavior in partially observable environments; and integrated control paradigms for mobile manipulators. Our approach advocates actively gathering salient information, modeling the environment, reasoning about solutions to new problems, and coordinating ad hoc interactions between multiple degrees of freedom to do mechanical work. Representations that encode control knowledge are a primary concern. Individual robots must exploit declarative structure for planning and must learn procedural strategies that work in recognizable contexts. We present several pieces of an overall framework in which a robot learns situated policies for control that exploit existing control knowledge and extend its scope. Several examples drawn from the research agenda at the Laboratory for Perceptual Robotics are used to illustrate the ideas.

In this paper, we describe the advances in the design, actuation, modeling, and control field of continuum robots. After decades of pioneering research, many innovative structural design and actuation methods have arisen. Untethered magnetic robots are a good example; its external actuation characteristic allows for miniaturization, and they have gotten a lot of interest from academics. Furthermore, continuum robots with proprioceptive abilities are also studied. In modeling, modeling approaches based on continuum mechanics and geometric shaping hypothesis have made significant progress after years of research. Geometric exact continuum mechanics yields apparent computing efficiency via discrete modeling when combined with numerical analytic methods such that many effective model-based control methods have been realized. In the control, closed-loop and hybrid control methods offer great accuracy and resilience of motion control when combined with sensor feedback information. On the other hand, the advancement of machine learning has made modeling and control of continuum robots easier. The data-driven modeling technique simplifies modeling and improves anti-interference and generalization abilities. This paper discusses the current development and challenges of continuum robots in the above fields and provides prospects for the future.

The soft robot is a new type of robot with strong adaptability, good pliability, and high flexibility. Today, it is widely used in the fields of bioengineering, disaster rescue, industrial production, medical services, exploration, and surveying. In this paper, the typical driven methods, 3D printing technologies, applications, the existed problems, and the development prospects for soft robots are summarized comprehensively. Firstly, the driven methods and materials of the soft robot are introduced, including fluid driven, smart materials driven, chemical reaction driven, a twisted and coiled polymer actuator, and so on. Secondly, the basic principles and characteristics of mainstream 3D printing technologies for soft materials are introduced, including FDM, DIW, IP, SLA, SLS, and so on. Then, current applications of soft robots, such as bionic structures, gripping operations, and medical rehabilitation are described. Finally, the problems existing in the development of soft robots, such as the shortage of 3D printable soft materials, efficient and effective manufacturing of soft robots, shortage of smart soft materials, efficient use of energy, the realization of complex motion forms of soft robot, control action accuracy and actual kinematic modeling are summarized. Based on the above, some suggestions are put forward pertinently, and the future development and applications of the soft robot are prospected.

In this paper, we presents a novel approach for tracking and catching operation of space robots using learning and transferring human control strategies (HCS). We firstly use an efficient support vector machine (SVM) to parametrize the model of HCS. Then we develop a new SVM-based learning structure to better implement human control strategy learning in tracking and capturing control. The approach is fundamentally valuable in dealing with some problems such as small sample data and local minima, and so on. Therefore this approach is efficient in modeling, understanding and transferring its learning process. The simulation results attest that this approach is useful and feasible in generating tracking trajectory and catching objects autonomously.

SUMMARYIn recent years, soft robotics is widely considered as the most promising field for both research and application. First of all, the actuator is fundamental for designing, modeling, and controlling of soft robots. This paper presents a new type of pneumatic trunk-like soft actuator, which contains a chamber for stiffness adjustment in addition to three chambers for driving. Thus, the salient feature of the proposed actuator is the ability of stiffness self-regulation. The structure of the proposed actuator is described in detail. Then the theoretical models for elongation and bending motion of the actuator are established. The elongation as well as single-chamber and multi-chamber driving bending of the actuator were tested to verify the mathematical models. Finally, a dual-segment soft robot based on the proposed trunk-like soft actuator was developed and tested by experiments, which implies its potential application in practice.

In this paper, we address the problem of how to model human real-time control strategy and how to transfer that model to robots or humans. This class of problems is significant to a number of research areas, such as the Intelligent Vehicle Highway System, human-machine interfacing, space telerobotics, and virtual reality. Human models can benefit not only the development of more intelligent control strategies for robots and machines, but can also improve the transfer of human intelligence and skill from expert to apprentice. In this paper, we illustrate a system we developed for modeling human control strategy through the use of flexible cascade neural networks, which adjust the size of the network as part of the training process, and which can be extended with variable activation functions and node-decoupled extended Kalman filtering to achieve faster learning and better error convergence. We implement the method in modeling human real-time driving strategy and show that the HCS models converge to stable behavior, while preserving the differences between individuals’ varying control strategies. We discuss the use of HCS models for transferring skill from human expert to human apprentice; rather than learn directly from a human expert, a HCS model serves as a virtual teacher to a learning apprentice. Finally, we outline on-going research issues and future work related to human control strategy modeling and transfer, including stochastic model validation, and HCS model input selection.

The evolution of nature created delicate structures and organisms. With the advancement of technology, especially the rise of additive manufacturing, bionics has gradually become a popular research field. Recently, researchers have concentrated on soft robotics, which can mimic the complex movements of animals by allowing continuous and often responsive local deformations. These properties give soft robots advantages in terms of integration and control with human tissue. The rise of additive manufacturing technologies and soft matters makes the fabrication of soft robots with complex functions such as bending, twisting, intricate 3D motion, grasping, and stretching possible. In this paper, the advantages and disadvantages of the additive manufacturing process, including fused deposition modeling, direct ink writing, inkjet printing, stereolithography, and selective laser sintering, are discussed. The applications of 3D printed soft matter in bionics, soft robotics, flexible electronics, and biomedical engineering are reviewed.

Compared with rigid robots, soft robots have better adaptability to the environment because of their pliability. However, due to the lower structural stiffness of the soft manipulator, the posture of the manipulator is usually decided by the weight and the external load under operating conditions. Therefore, it is necessary to conduct dynamics modeling and movement analysis of the soft manipulator. In this paper, a fabric reinforced soft manipulator driven by a water hydraulic system is firstly proposed, and the dynamics of both the soft manipulator and hydraulic system are considered. Specifically, a dynamic model of the soft manipulator is established based on an improved Newton–Euler iterative method, which comprehensively considers the influence of inertial force, elastic force, damping force, as well as combined bending and torsion moments. The dynamics of the water hydraulic system consider the effects of cylinder inertia, friction, and water response. Finally, the accuracy of the proposed dynamic model is verified by comparing the simulation results with the experimental data about the steady and dynamic characteristics of the soft manipulator under various conditions. The results show that the maximum sectional error is about 0.0245 m and that the maximum cumulative error is 0.042 m, which validate the effectiveness of the proposed model.

In recent years, there has been a lot of research using reinforcement learning algorithms to train 2-legged robots to move, but there are still many challenges. The authors propose the GCTD3 method, which takes the idea of using Graph Convolutional Networks to represent the kinematic link features of the robot, and combines this with the Twin-Delayed Deep Deterministic Policy Gradient algorithm to train the robot to move. Graph Convolutional Networks are very effective in graph-structured problems such as the connection of the joints of the human-like robots. The GCTD3 method shows better results on the motion trajectories of the bipedal robot joints compared with other reinforcement learning algorithms such as Twin-Delayed Deep Deterministic Policy Gradient, Deep Deterministic Policy Gradient and Soft Actor Critic. This research is implemented on a 2-legged robot model with six independent joint coordinates through the Robot Operating System and Gazebo simulator.

In a step toward soft robot proprioception, and therefore better control, this paper presents an internally illuminated elastomer foam that has been trained to detect its own deformation through machine learning techniques. Optical fibers transmitted light into the foam and simultaneously received diffuse waves from internal reflection. The diffuse reflected light was interpreted by machine learning techniques to predict whether the foam was twisted clockwise, twisted counterclockwise, bent up, or bent down. Machine learning techniques were also used to predict the magnitude of the deformation type. On new data points, the model predicted the type of deformation with 100% accuracy and the magnitude of the deformation with a mean absolute error of 0.06°. This capability may impart soft robots with more complete proprioception, enabling them to be reliably controlled and responsive to external stimuli.

Agricultural robots are one of the important means to promote agricultural modernization and improve agricultural efficiency. With the development of artificial intelligence technology and the maturity of Internet of Things (IoT) technology, people put forward higher requirements for the intelligence of robots. Agricultural robots must have intelligent control functions in agricultural scenarios and be able to autonomously decide paths to complete agricultural tasks. In response to this requirement, this paper proposes a Residual-like Soft Actor Critic (R-SAC) algorithm for agricultural scenarios to realize safe obstacle avoidance and intelligent path planning of robots. In addition, in order to alleviate the time-consuming problem of exploration process of reinforcement learning, this paper proposes an offline expert experience pre-training method, which improves the training efficiency of reinforcement learning. Moreover, this paper optimizes the reward mechanism of the algorithm by using multi-step TD-error, which solves the probable dilemma during training. Experiments verify that our proposed method has stable performance in both static and dynamic obstacle environments, and is superior to other reinforcement learning algorithms. It is a stable and efficient path planning method and has visible application potential in agricultural robots.

One of the main challenges in robotics is the development of systems that can adapt to their environment and achieve autonomous behavior. Current approaches typically aim to achieve this by increasing the complexity of the centralized controller by, e.g., direct modeling of their behavior, or implementing machine learning. In contrast, we simplify the controller using a decentralized and modular approach, with the aim of finding specific requirements needed for a robust and scalable learning strategy in robots. To achieve this, we conducted experiments and simulations on a specific robotic platform assembled from identical autonomous units that continuously sense their environment and react to it. By letting each unit adapt its behavior independently using a basic Monte Carlo scheme, the assembled system is able to learn and maintain optimal behavior in a dynamic environment as long as its memory is representative of the current environment, even when incurring damage. We show that the physical connection between the units is enough to achieve learning, and no additional communication or centralized information is required. As a result, such a distributed learning approach can be easily scaled to larger assemblies, blurring the boundaries between materials and robots, paving the way for a new class of modular “robotic matter” that can autonomously learn to thrive in dynamic or unfamiliar situations, for example, encountered by soft robots or self-assembled (micro)robots in various environments spanning from the medical realm to space explorations.

A neuroadaptive control scheme for elastic-joint robots is proposed that uses a relatively small neural network. Stability is achieved through standard Lyapunov techniques. For added performance, robust modifications are made to both the control law and the weight update law to compensate for only approximate learning of the dynamics. The estimate of the modeling error used in the robust terms is taken directly from the error of the network in modeling the dynamics at the currant state. The neural network used is the CMAC-RBF Associative Memory (CRAM), which is a modification of Albus's CMAC network and can be used for robots with elastic degrees of freedom. This results in a scheme that is computationally practical and results in good performance.

Study: Soft robots can achieve the desired range of motion for finger movement to match their axis of rotation with the axis of rotation of the human hand. The iterative design has been used to achieve data that makes the movement smooth and the range of movement wider, and the validity of the design has been confirmed through practical experiments. Limitation: The challenges facing this research are to reach the most significant inclined angle and increase the force generated by the actuator, which is the most complicated matter while maintaining the desired control accuracy. The motion capture system verifies the actual movement of the soft pneumatic actuator (SPA). A tracking system has been developed for SPA in action by having sensors to know the position and strength of the SPA. Results: The novelty of this research is that it gave better control of soft robots by selecting the proportional, integral, and derivative (PID) controller. The parameters were tuned using three different methods: ZN (Ziegler Nichols Method), GA (Genetic Algorism), and PSO (Particle Swarm Optimization). The optimization techniques were used in Methods 2 and 3 in order to reach the nominal error rate (0.6) and minimum overshoot (0.1%) in the shortest time (2.5 s). Impact: The effect of the proposed system in this study is to provide precise control of the actuator, which helps in medical and industrial applications, the most important of which are the transfer of things from one place to another and the process of medical rehabilitation for patients with muscular dystrophy. A doctor who treats finger muscle insufficiency can monitor a patient’s ability to reach a greater angle of flexion or increase strength by developing three treatment modalities to boost strength: Full Assisted Movement (FAM), Half Assisted Movement (HAM), and Resistance Movement (RM).

Spiders rely on a hydraulic system to stretch their legs but use muscles to make their legs flex. The compound drive of hydraulics and muscle makes an integrate dexterous structure with powerful locomotion abilities, which perfectly meets the primary requirements of advanced robots. Inspired by this hydraulics-muscle co-drive joint, a novel flexible joint actuator was proposed and its driving characteristics were preliminarily explored. The bio-inspired flexible joint manifested as a double-constrained balloon actuator, which was fabricated by the composite process of 3D printing and casting. To evaluate its performance, the mathematical model was deduced, as well as the finite element analysis (FEA) model. A series of experiments on the rotation angles, driving forces, and efficiencies of the flexible joint were carried out and compared with the mathematical calculations and FEA simulations. The results show that the accuracy of the two theoretical models can be used to assess the joint actuator. The locomotion test of a soft arthropod robot with two flexible joints was also implemented, where the moving speed reached 22 mm/s and the feasibility of the proposed flexible joint applied to a soft robot was demonstrated.

The "information revolution" of our time affects our entire generation. While a vision of the "Information Society," with its financial, legal, business, privacy, and other aspects has emerged in the past few years, the "traditional scene" of information technology, that is, industrial automation, maintained its significance as a field of unceasing development. Since the old-fashioned concept of "Hard Automation" applicable only to industrial processes of fixed, repetitive nature and manufacturing large batches of the same product1)was thrust to the background by keen market competition, the key element of this development remained the improvement of "Machine Intelligence". In spite of the fact that L. A. Zadeh already introduced the concept of "Machine Intelligence Quotient" in 1996 to measure machine intelligence2) , this term remained more or less of a mysterious meaning best explicable on the basis of practical needs. The weak point of hard automation is that the system configuration and operations are fixed and cannot be changed without incurring considerable cost and downtime. Mainly it can be used in applications that call for fast and accurate operation in large batch production. Whenever a variety of products must be manufactured in small batches and consequently the work-cells of a production line should be quickly reconfigured to accommodate a change in products, hard automation becomes inefficient and fails due to economic reasons. In these cases, new, more flexible way of automation, so-called "Soft Automation," are expedient and suitable. The most important "ingredient" of soft automation is its adaptive ability for efficiently coping with changing, unexpected or previously unknown conditions, and working with a high degree of uncertainty and imprecision since in practice increasing precision can be very costly. This adaptation must be realized without or within limited human interference: this is one essential component of machine intelligence. Another important factor is that engineering practice often must deal with complex systems of multiple variable and multiple parameter models almost always with strong nonlinear coupling. Conventional analysis-based approaches for describing and predicting the behavior of such systems in many cases are doomed to failure from the outset, even in the phase of the construction of a more or less appropriate mathematical model. These approaches normally are too categorical in the sense that in the name of "modeling accuracy," they try to describe all structural details of the real physical system to be modeled. This significantly increases the intricacy of the model and may result in huge computational burden without considerably improving precision. The best paradigm exemplifying this situation may be the classic perturbation theory: the less significant the achievable correction is, the more work must be invested for obtaining it. Another important component of machine intelligence is a kind of "structural uniformity" giving room and possibility to model arbitrary particular details a priori not specified and unknown. This idea is similar to that of the ready-to-wear industry, whose products can later be slightly modified in contrast to the custom-tailors' made-to-measure creations aiming at maximum accuracy from the beginning. Machines carry out these later corrections automatically. This "learning ability" is another key element of machine intelligence. To realize the above philosophy in a mathematically correct way, L. A. Zadeh separated Hard Computing from Soft Computing. This revelation immediately resulted in distinguishing between two essential complementary branches of machine intelligence: Hard Computing based Artificial Intelligence and Soft Computing based Computational Intelligence. In the last decades, it became generally known that fuzzy logic, artificial neural networks, and probabilistic reasoning based Soft Computing is a fruitful orientation in designing intelligent systems. Moreover, it became generally accepted that soft computing rather than hard computing should be viewed as the foundation of real machine intelligence via exploiting the tolerance for imprecision, uncertainty and partial truth to achieve tractability, robustness, low solution cost and better rapport with reality. Further research in the past decade confirmed the view that typical components of present soft computing such as fuzzy logic, neurocomputing, evolutionary computation and probabilistic reasoning are complementary and best results can be obtained by their combined application. These complementary branches of Machine Intelligence, Artificial Intelligence and Computational Intelligence, serve as the basis of Intelligent Engineering Systems. The huge number of scientific results published in journals and conference proceedings worldwide substantiates this statement. Three years ago, a new series of conferences in this direction was initiated and launched with the support of several organizations including the IEEE Industrial Electronics Society and IEEE Hungary Section in technical cooperation with IEEE Robotics & Automation Society. The first event of the series hosted by Bdnki Dondt Polytechnic, Budapest, Hungary, was called "19997 IEEE International Conference on Intelligent Engineering Systems " (INES'97). The Technical University of Vienna, Austria hosted the next event of the series in 1998, followed by INES'99 held by the Technical University of Kosice, Slovakia. The present special issue consists of the extended and revised version of the most interesting papers selected out of the presentations of this conference. The papers exemplify recent development trends of intelligent engineering systems. The first paper pertains to the wider class of neural network applications. It is an interesting report of applying a special Adaptive Resonance Theory network for identifying objects in multispectral images. It is called "Extended Gaussian ARTMAP". The authors conclude that this network is especially advantageous for classification of large, low dimensional data sets. The second paper's subject belongs to the realm of fuzzy systems. It reports successful application of fundamental similarity relations in diagnostic systems. As an example failure detection of rolling-mill transmission is considered. The next paper represents the AI-branch of machine intelligence. The paper is a report on an EU-funded project focusing on the storage of knowledge in a corporate organizational memory used for storing and retrieving knowledge chunks for it. The flexible structure of the system makes it possible to adopt it to different SMEs via using company-specific conceptual terms rather than traditional keywords. The fourth selected paper's contribution is to the field of knowledge discovery. For this purpose in the first step, cluster analysis is done. The method is found to be helpful whenever little or no information on the characteristics of a given data set is available. The next paper approaches scheduling problems by the application of the multiagent system. It is concluded that due to the great number of interactions between components, MAS seems to be well suited for manufacturing scheduling problems. The sixth selected paper's topic is emerging intelligent technologies in computer-aided engineering. It discusses key issues of CAD/CAM technology of our days. The conclusion is that further development of CAD/CAM methods probably will serve companies on the competitive edge. The seventh paper of the selection is a report on seeking a special tradeoff between classical analytical modeling and traditional soft computing. It nonconventionally integrates uniform structures obtained from Lagrangian Classical Mechanics with other simple elements of machine intelligence such as saturated sigmoid transition functions borrowed from neural nets, and fuzzy rules with classical PID/ST, and a simplified version of regression analysis. It is concluded that these different components can successfully cooperate in adaptive robot control. The last paper focuses on the complexity problem of fuzzy and neural network approaches. A fuzzy rule base, be it generated from expert operators or by some learning or identification schemes, may contain redundant, weakly contributing, or outright inconsistent components. Moreover, in pursuit of good approximation, one may be tempted to overly assign the number of antecedent sets, thereby resulting in large fuzzy rule bases and much problems in computation time and storage space. Engineers using neural networks have to face the same complexity problem with the number of neurons and layers. A fuzzy rule base and neural network design, hence, have two important objectives. One is to achieve a good approximation. The other is to reduce the complexity. The main difficulty is that these two objectives are contradictory. A formal approach to extracting the more pertinent elements of a given rule set or neurons is, hence, highly desirable. The last paper is an attempt in this direction. References 1)C. W. De Silva. Automation Intelligence. Engineering Application of Artificial Intelligence. Vol. 7. No. 5. 471-477 (1994). 2)L. A. Zadeh. Fuzzy Logic, Neural Networks and Soft Computing. NATO Advanced Studies Institute on Soft Computing and Its Application. Antalya, Turkey. (1996). 3)L. A. Zadeh. Berkeley Initiative in Soft Computing. IEEE Industrial Electronics Society Newsletter. 41, (3), 8-10 (1994).

This article presents anomaly detection algorithms for marine robots based on their trajectories under the influence of unknown ocean flow. A learning algorithm identifies the flow field and estimates the through-water speed of a marine robot. By comparing the through-water speed with a nominal speed range, the algorithm is able to detect anomalies causing unusual speed changes. The identified ocean flow field is used to eliminate false alarms, where an abnormal trajectory may be caused by unexpected flow. The convergence of the algorithms is justified through the theory of adaptive control. The proposed strategy is robust to speed constraints and inaccurate flow modeling. Experimental results are collected on an indoor testbed formed by the Georgia Tech Miniature Autonomous Blimp and Georgia Tech Wind Measuring Robot, while simulation study is performed for ocean flow field. Data collected in both studies confirm the effectiveness of the algorithms in identifying the through-water speed and the detection of speed anomalies while avoiding false alarms.