Formation control of nonholonomic mobile robots and extended algorithms

点击翻译按钮获取中文摘要

We have developed following three topics. (a)Control algorithm for organizing specified formation by nonholonomic mobile robots, which considers collision avoidance among robots., (b)Motion control algorithm for goal-reaching task by the formed robot group, (c)Control and formation organization of nonholonomic mobile robots in 3D space.To organize desired formation among nonholonomic mobile robots, we employee exact linearization of nonlinear target system and back-stepping method to develop control input. Additionally, to avoid collisions among mobile robots, the proposed control strategy is composed three control stages ; virtual robot tracking control (VR control), distance control among two robots (l-l control), and that for three robots (l・ψ control). By developing each control low and designing transition strategy, we have realized the objective to form a desired formation considering collision avoidance.When robots move in a formation, we have to consider collisions not only amo … More ng robots but also with obstacles in an environment. If the robots exactly keep their formation, their mobility in actual environment decreases because the size of the robots is enlarged comparing with a single one. To relax the problem, we developed motion algorithm that the robots changes their formation according to the environment. We employee algorithms of computational geometry and potential-like control low. By the algorithm, the robot forms stable formation in free-space and snake-like one in obstacle regions.We have extended the algorithms for a nonholonomic robot group in 3D space. We have considered formation keeping and motion control for a group of Unmanned Underwater Autonomous Vehicles(UAV). We also developed an algorithm to extract important features to avoid collisions with complex submarine topology by means of Delaunay tessellation in computational geometry.We have verified the effectiveness of the proposed methods by carrying out computer simulations and experiments with maximum 6 car-like small robots, which have nonholonomic constraints. Less