Abstract: The perception and representation of unknown complex terrains restrict the raise of motion property and operational efficiency of hexapod robots for a long time. Aiming at the significant issues of traditional terrain perception and representation methods based on external sensing, such as the limitation of perception range, insufficient perception accuracy and poor representation effect, the method of perception and representation of roaming terrain for hexapod robot based on foot positions is innovatively proposed by researching on the terrain perception mechanism of multi-legged creatures and making full use of the characteristic of alternative discrete contact between feet and terrains. A computing model is constructed to efficiently obtain the coordinates of disordered foot positions in the time-varying body coordinate system. Based on the periodic processing and vectorized description of foot sequences, a local terrain representation method based on foot positions is established and the mapping relationship between the time-varying body poses and local terrain is constructed. The coupling constraint and the transformation mechanism between body poses during adjacent periods is systematically analyzed and then the global terrain reconstruction method is established. The splicing of periodic local mapping terrain by setting continuous and accurate body poses as the reference is realized. The experiment shows that this method can represent different local terrains more accurately and reasonably without additional external sensors than traditional methods and accurately reconstruct the global topography of roaming terrain.

Key words: hexapod robot, kinematics analysis, terrain perception, quantitative characterization, morphology reconstruction