六足机器人非结构化地形下步态生成与腿部控制研究-机械电子工程专业论文.docxVIP

哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 万方数据 万方数据 Abstract Hexapod robot has a strong ability to adapt to unstructured terrain due to the redundancy of its leg structure, characteristics of time-varying topology and discreteness of its footholds selection. Footholds selection, being an important part during of the movement of hexapod robot, its reasonability is closely related to the moving stability, motion speed, and overstepping ability of the robot. Therefore, studying the gait generation of hexapod robot in unstructured terrains has great important theoretical significance and practical value. In this paper, firstly the structure of the hexapod robot is described, and kinematics of swing leg and workspace of the foot-end are analyzed from the perspective of serial robot based on the theory of robotics, as well as the kinematics of torso and torso workspace from the view of the parallel robot. Afterwards, the simulation results of ADAMS and the computed results of MATLAB are compared to verify the analysis of kinematics of series-parallel structure. To have the maximum flexibility in movement, the initial state of the hexapod robot is optimized using Multi-Objective Partical Swarm Algorithm. Based on three-legged gait, the footstep of hexapod robot is planned combining with the known terrain information, i.e., gait generation. The small-scale fast planning strategies is proposed to select the drop-foot point for rapidity and effectiveness. A series of penalty functions can be generated by analyzing terrain constraints, motion constraints of robot and stability constraints needed to be considered in the select of the drop-foot point. Comprehensively considering the penalty function and using them to evaluate the foot drop point, the footholds plan in this paper is transformed into an optimization problem which is solved by Ant Colony Algorithm. Finally, MATLAB simulation is carried out to verify the effectiveness of the footholds plan. After fully considering the ability