A Study Of Locomotion And Force Planning For Multilegged Walking Robots

Multilegged walking robots have been generating a considerable interest, because of their high performances in various robotic tasks and their great mobility and adaptability to the rough terrain. The research on the robots is of momentous scientific significance and practical application value. This thesis addresses kinematics, dynamics and force planning, stability and gait generation for multilegged walking robots. In order to solve these problems, a systematic study for the robots is presented in the thesis. The research work in this thesis includes following. Firstly, the relationships of the position and velocity between the independent actuation joints and the redundant ones of a multilegged robot are derived by analyzing the natural constraints of the robot mechanism. Based on the work, a whole kinematics of crawl locomotion is studied for the walking robots treated as an overall kinematic chain, which includes the inverse and forwards processes of position, velocity and acceleration. Secondly, the dynamic model of the walking locomotion is derived on the basis of kinematics analysis. The locomotion and force constraints are formulated and solved. Further, by transforming the friction constraints from the nonlinear inequalities into a combination of linear equalities and linear inequalities, by eliminating the linear equality constraints from the original problem, and by solving a quadratic optimization problem, a new method for optimal force distribution for legs of the walking robot is presented. Thirdly, a novel approach to static stability analysis of the robot is shown by using the space relationships of the locomotion mechanism and by consequentially introducing a new concept of the Statically Stable Area for footholds of the robot in any body attitude. The stability formulation can be used to generate gait for the robots. The maximized stride for the translational gaits and various standstill-turning gaits is determined by considering the coordinated locomotion of the legs. The generalized and explicit formulation for the automatic generation of translational gaits and various standstill-turning gaits is then discussed. Based on the above work, an efficient gait algorithm is developed for the omnidirectional and continuous crawl of the robots. Fourthly, the experiments of gait generation and control are conducted in a multilegged walking robot developed at our lab. The experimental results verify the correctness and effectiveness of the work in this thesis. The research in this thesis considerably extends and deepens the existing work concerned for multilegged walking robots. The major contributions of the research are as follows: 1) This thesis extends the crawl control of multilegged walking robots by introducing the kinematics of the locomotion for the robots with a kinematic topology and redundant actuation. 2) A new static stability explicitly formulated, which is superior to the existing approaches. Based on the above work, the gait algorithm is developed to implement the omnidirectional and continuous crawl for the robots. The proposed gait outperforms the conventional gaits in natural transition, possible big stride, and travel mobility. And 3) The dynamic model of the walking locomotion is built, and a novel optimal force distribution for legs of a robot is presented. The technique is compared with the conventional methods to show its superior performance in terms of size of problem, quality of solution, and scope of application.