Terrain Modeling and Motion Planning for Hexapod Walking Robot Control

A planning technique for six-legged walking robot is presented in this thesis along with a terrain modelling approach that allows to exploit possible supporting footholds of the surface. The proposed approach for merging measurements provided by a depth camera sensor allows to reconstruct the robot surrounding. Based on the measurements of the environment, a grid based elevation map is constructed to create a suitable world model supporting motion planning for the six-legged walking robot. Such a model is adjusted to precisely represent a problematic passage in front of the robot by that provides only a limited number of possible locations for the robot footholds. Furthermore, an approach for an efficient utilization of the suitable positions of the terrain is proposed based on an evaluation function. Practical simplifications of the robot motion representation are proposed to efficiently speed up the planning of the robot forward motion over the terrain. Moreover, the proposed approach was successfully verified in real experiments.