Research On Motion Planning Methods For Underactuated Robotic Systems And Their Applications

In recent years,the rapid developments of the mechanical design,materials science,electronic information,artificial intelligence,sensors and other related technologies have promoted development of the robotic technology.The robotic systems are not only conditional applied in traditional industrial field,but also extended to various aspects of our social life,such as home entertainments,medical services,equipments detection and etc.Facing to the task requirements of flexibility and new applicable scenarios with space and energy constraints,underactuated robotic systems have advantages on light weight,low energy consumption,high flexibility,strong robustness,which draw lots of attentions from researchers all over the world.In some difficult,high operating costs or high risk tasks,researchers hope that the underactuated robotic systems can work intelligently and autonomously to replace manual operation and improve operating accuracy and reliability.Therefore,today's underactuated robotic systems are generally equipped with advanced sensors,high-performance processors and high quality motion planning algorithms to work autonomously in novel complex scenarios.The thesis focuses on the motion planning method research.Considering to application backgrounds of the different underactuated robotic systems,new type of motion planning methods are designed to drive various underactuated robotic systems autonomous working and simultaneously to improve the quality and efficiency of operating performances.In particular,the main contributions of the thesis are listed as follows:(1)Aiming at the uncontrollable risks in current robot-assist surgery,such as fatigue,jitter and pose changes,when surgeon manipulates medical robot in small clinical environments for a long time,a three-dimensional backward average neural dynamic motion planning method is proposed for the tendon-driven serpentine surgical robotic system.In this method,the sweeping area of the whole robot movement is regarded as the main influencing factor during motion planning computing.As a result,unnecessary sweeping area during the tendon-driven serpentine surgical robot movement is reduced as well as the risk of accidents.Then,the autonomy and safety of the surgical robot are guaranteed.(2)Facing to the requirement of the wheeled service mobile robot autonomous working in unknown dynamic environments,an improved geometrical learning motion planning algorithm is proposed based on its traditional version.This method considers the nonholonomic constraint of the wheeled service mobile robot,optimizes the random selection strategy,improves the poor convergence of the traditional geometric learning motion planning algorithm when the robot is close to the target,and designs an adaptive velocity moving strategy.Simulations and experiments show that the proposed method has better planning performance than its traditional version.Meanwhile,the computational efficiency of planning method is improved that can meet the needs of wheeled service mobile robot autonomous working in an indoor environment.(3)Considering to the problem of the computational exponential explosion when traditional motion planning methods solving high dimensional planning problem of the complex robotic systems,an incremental sampling-based motion planning method based on local environments is proposed.This method improves the collision checking procedure by referred an idea of establishing an environmental information model for obstacles.In particular,we build a probabilistic collision risk for detected regions,which is updated and extended as environments changes.Then,by designing a novel cost function consisting of distance information and probabilistic collision risk,the motion planning algorithm does not need to execute collision checking at the sampling stage.As a benefit,the convergence speed and computational efficiency are promoted.The comparative simulation results verify the effectiveness and superiority of the proposed method.Furthermore,the proposed method is applied to self-developed ground wheeled mobile robots and micro quadrocopter system for validation respectively.The results show the incremental sampling-based motion planning method based on local environments has better performance than the RRT and the RRT*,which also provide a guidance for its further applications in other more complex robotic systems.Finally,the main contributions and innovations of the thesis are summarized and the further research works are prospected.