Guest Editorial: Autonomous systems: Navigation, learning, and control
Yu Zhang, Fei Gao, Yuxiang Sun, Naira Hovakimyan, Zheng Fang · IET Cyber-Systems and Robotics · 2021
This is the IET Cyber-systems and Robotics special issue of Autonomous systems: Navigation, learning, and control. Autonomous systems, as very important representatives of Artificial Intelligence technologies, combine mechanical and electronic hardware, operating system, low-level dynamic control, and high-level intelligent decision components to address challenges that demand high-level autonomy and machine intelligence. Autonomous systems such as aerial robotics, ground vehicles, unmanned surface vehicles, and even all-terrain vehicles for aerospace applications, etc., have played an essential role in many aspects. For example, during the COVID-19 pandemic, autonomous systems have been used for disinfection and food delivery to reduce infection risks. However, navigation, learning, and control are critical for realizing true autonomy, which are still left for research. Navigation is a critical technology that works as the high-level intelligent decision component and directly determines the efficiency of conducting autonomous tasks. Learning, which is regarded as the most representative technology of artificial intelligence, has become a dominating method in many fields, such as image processing and understanding. There are also trends of adopting learning-based methods on autonomous robots, which are the perfect field to adopt state-of-the-art artificial intelligence technologies. The low-level dynamic control part, which builds the foundation of autonomous tasks, receives long-term research interest from modern control theory to intelligent control theory. So in this special issue, these three critical elements are examined to bring valuable inspirations to the community. Seven papers in total are selected in the special issue covering the topic of navigation, learning, and control. Paper 1: RGB-D SLAM with moving object tracking in dynamic environments, Paper 2: Fast-Tracker 2.0: Improving autonomy of aerial tracking with active vision and human location regression, Paper 3: A survey of learning-based robot motion planning, Paper 4: Deep reinforcement learning for shared control of mobile robots, Paper 5: Predictive-based optimal automatic formation control of mobile vehicles, Paper 6: Technical report: PID design of second-order non-linear uncertain systems with fractional order operations, Paper 7: Control of upper limb rehabilitation robot based on active disturbance rejection control. Paper 1 by Weichen Dai et al. investigates the SLAM problem in the dynamic environment, whereas Paper 2 by Neng Pan et al. studies the autonomous tracking problem. Both are essential aspects of navigation. The SLAM method can help improve the robustness of localization, enabling navigation tasks in many harsh environments. Also, the autonomous tracking problem is a vivid application of task-driven navigation technology. In Paper 1, Weichen Dai et al. exploit the clue of point correlation to distinguish the moving objects without any prior knowledge, achieving SLAM and tracking at the same time. Their method only adopts the static points to conduct pose estimation and utilizes the unstable points to track the moving objects. Besides, a volume overlap checking step is designed to merge multiple groups of dynamic points that belong to the same moving object. In Paper 2, Neng Pan et al. extend their prior work by introducing a novel framework based on deep learning and non-linear regression to track the human target, addressing the drawbacks of their previous work. Besides, the trajectory planning part is also improved by introducing an occlusion-aware mechanism. Their experiment results indicate that their upgrade helps improve the robustness when handling challenging tasks. Recent years, the growing interest is witnessed in adopting learning-based methods in challenging motion planning tasks. Paper 3 by Jiankun Wang et al. presents a comprehensive survey of learning-based robot motion planning. Three types of learning methods, including supervised, unsupervised, and reinforcement learning, are addressed in this survey. The survey provides the classical and learning-related definitions of motion planning problems, along with the application of learning-based algorithms. The survey also points out some potential ways to combine classical motion-planning and learning techniques. Combing learning and control has shown promising results recently. In Paper 4, Chong Tian et al. adopt deep reinforcement learning to tackle the problem of shared control for mobile robots. They develop an extended Twin Delayed DDPG based shared control framework with the Reinforcement Learning agent to give the optimal shared control ratio, which can learn to assist the human operator. They also develop an extended DAGGER human agent for training the RL agent. Control, as the low-level basis of autonomous systems, has long been actively researched. In Paper 5, Tadanao Zanma et al. study the optimal automatic formation control to deal with the control problem of multiple mobile robots systems. They develop a novel predictive-based automatic formation control method, where the unique formation is derived from feasible formations, and the control is built by mixed-integer quadratic programing (MIQP). Paper 6 by Song Chen et al. demonstrates a technical report on designing a PID controller for second-order non-linear uncertain systems with fractional order operations. They demonstrate the stability verification of the proposed control system along with the simulation results, indicating the proposed system achieved global stabilization under some suitable conditions. Paper 7 by Junchen Li et al. discusses the control of upper-limb rehabilitation robots using active disturbance rejection control (ADRC) to resolve the problems of slow convergence speed and poor tracking accuracy. They propose a novel algorithm based on ADRC, which has better trajectory-tracking performance and quick convergence speed. They also prove the convergence of the main structure. All of the papers selected for this special issue indicate the emerging research trends in the field of autonomous systems regarding navigation, learning, and control. We hope this special issue can benefit researchers worldwide by communicating and sharing the latest works. The Guest Editors wish to express their gratitude to IET Cyber-systems and Robotics Editors-in-Chief Rob Buckingham, Jian Chu, and Max Meng for their constant support. Finally, they are grateful to the colleagues who agreed to provide reviews for the special issue and sent the valuable comments. The authors declare no conflict of interests.