Research On Gait Planning And Adaptive Control For Multi-legged Robot

Multi-legged robot has very strong adaptability to the environment and flexible movement with its relatively simple structure and control methods, steadily walking, redundant physical. All these make it being paid attention by the researchers from all over the world and become a remarkable research field in research of robot. But the multi-legged feature of the robot leads to the complexity of motion control at the same time. Therefore, gait planning and control has become a key technology to the development of Multi-legged robot. In this thesis, multi-legged robot’s gait planning and adaptive control have been studied. The thesis is mainly divided into the following areas:In terms of mechanical structure and control systems of Multi-legged robot, three degrees of freedom for the legs institutions is designed and hierarchical control system is established for multi-legged robot’s mechanical structure and control systems, which made the robot has good moving performance.In terms of motion control of Multi-legged robot, a CPG model with fast self-oscillation is designed to overcome the deficiency of the traditional CPG model. Based on that, the parameter characteristics and motion states characteristic of CPG model are studied through the computer numerical simulation in order to make the multi-legged robot walk steadily by the control of CPG.CPG model is used to control multi-legs coordination for Multi-legged robot. And the CPG control model for Multi-legged robot is simplified through building the motion mapping function. For the problem of the complexity of the CPG parameter characteristics and the difficulty of the adjustment in this thesis, the genetic is put into use to optimize it, the typical gaits of the Multi-legged robot have been analyzed and experimented, meanwhile. The gait planning method based on CPG is proved reasonably.In order to improve the adaptive capacity of Multi-legged robot in unstructured and uncertain environment, feedback control and CPG model are combined in thesis. Then stretch reflexes, posture reflection and obstacles reflection are modeling respectively. Stable moving and obstacle avoidance movement have been achieve at the Multi-legged robot, and then physical prototype Multi-legged robot experiments are used to demonstrate that the design of control methods is reliable which meet the design requirements.