With the extension of the application range of quadruped mobile robot, the more and more researches have been done. However, in the complex environment, there still have some works to do on how to keep balance and stability. In this thesis, a quadruped robot with parallel-serial leg mechanism to work in more structural and unstructured environments is posed in order to avoid failure of sensors and actuator due to environmental factors.Firstly, Structure analyzes and kinematics model is the basis of gait planning. In this thesis, the DOFs of one swing leg, three-legs touchdown and four-legs touchdown are calculated. Then kinematics analyzes of leg and the transformation between each local coordinate and global coordinate are given. The actual work region of toe is selected from full workspace. Three-legs touchdown and four-legs touchdown are equivalent to 3-UPS and 4-UPS parallel mechanism by chain equivalent method so that the inverse kinematics of original mechanism is given.Secondly, to improve adaptability of unstructured environment recognition method, a new method which is based on force feedback is presented, and full statics model of a leg is given. To identify simply environment, a discrete recognition method by installing contingency tablets in the bars and to complex environment, a continuous recognition method by inversely solving the force of toe from the stress of bar are introduced. The force acting on body is solved through static mechanical balance equations.Finally, the best gait planning of walking stably is demonstrated in energy consumption, efficiency and workspace on the situation of stability. With the constraints of workspace, the original orientation is discussed. Static stable gait planning is the most basic gait of walking stably. For the purpose of lager stable margin, the static gait of fixed center of gravity is planed prior, on this basis; the following static gait is planed for moving smoothly. At the same time, zero radius turning gait is planed to solve the problem of walking in different directions. Utilizing the proposed unstructured environment recognition method, methods of judging and escaping obstacles are analyzed and planed. |