We are pleased to announce the special issue on Recent Advances in Nonlinear Robot Control Technology.
Nonlinear control technology has played a key role since the inception of robot control. Beyond traditional nonlinearity in local robot dynamics, increasingly complex control systems require addressing nonlinearities at the system level. Recent groundbreaking advances in artificial intelligence have opened the door to transforming systems previously treated as just uncertainties, as typified by humans and environment, into nonlinear systems. It is no exaggeration to say that it is more difficult to find a system that is purely closed to the linear world. Despite high maturity of traditional nonlinear control, novel theories and control tools have continued to be developed, e.g. control barrier functions, optimal mass transport, formal methods, and data-driven and learning-based control. In this special issue, we have collected a representative body of innovative theoretical contributions that have potential applications to robotics as well as applicative robotics researches that show successful implementation of recent theory of nonlinear control.
Our special issue consists of two volumes. In this first volume, we highlight contributions related to learning-based control among a number of accepted papers. C.A. O’Hara and T. Yairi present a novel framework for optimizing energy efficiency and computational load in safety-critical robotic systems based on Graph Attention Networks for state awareness and decision-making. K. Kobayashi et al. propose a novel reward distribution mechanism for a surveillance system based on a multi-agent reinforcement learning method with an aggregator against a control specification described by a linear temporal logic formula. R. Wang et al. address the so-called optimal static output feedback control problem with measurement noise, and present a novel solution combining the density estimation and a state-of-the-art deep reinforcement learning method, Soft Actor-Critic. W. Zhijun et al. tackle self-triggered control for networked control systems having initially unknown system dynamics. The authors present a novel solution consisting of lifting techniques and lifted dynamics learning through Gaussian Process. P. Pongsing et al. address upper-limb rehabilitation, and present a position-based force control system based on Fuzzy-PI control, enabling a rehabilitation robot to autonomously generate self-adaptive resistance using electromyography signals from the patient. The experimental studies demonstrate that the proposed system outperforms a baseline method. The second volume will be published shortly.
We would like to thank all authors who submitted their original papers to the issue. We also would like to thank every reviewer who gave valuable and insightful comments all the way to improve the quality of each paper. Finally, we would like to show our appreciation to Ms. Noriko Watanabe of Advanced Robotics for her great assistance.