Locomotion over rough terrain is still a problem yet to be solved for legged robots. One of the problems arises from the inability to identify terrain roughness during locomotion, which could be crucial for decision-making and successful task completion. Our proposed terrain roughness method is inspired by the observation that humans can sense their limb position in space without looking at them, which allows us to estimate obstacle heights. This method is based on robot feet coordinate standard deviation (further referred to as SD) parameter evaluation. SD values could be categorized to represent different terrain roughness, and such categories could be useful for selecting different gaits for different terrains. In this paper, we investigate the possibility of using already known feet coordinates to evaluate terrain roughness by calculating their standard deviation (SD). We present simulation results that show that the SD value only depends on terrain roughness and is not influenced by large terrain slopes. Experiments were conducted with real robots while walking over obstacles with different gaits to validate the method. This research mainly aims to test how robot gaits influence SD parameters for terrain roughness evaluation. The experimental results showed that the SD parameter calculated from the robot’s foot coordinates can be used to evaluate terrain roughness. The robot’s gaits have little to no influence on the SD parameter.
Read full abstract