Robot Trotting with Segmented Legs in Simulation and Hardware.

This research is focusing on the implementation, testing, and analysis of quadrupedal, bio-inspired robot locomotion. Our tool of research is a light-weight, quadruped robot of the size of a house cat, both in simulation and hardware. We are currently following the idea of testing bio-inspired blue-prints such as leg-segmentation, directional leg compliance (bio-mechanical), and central pattern generators (bioinspired neuro-control) for their feasibility, and advantages against more traditional, engineered solutions. Clearly, our first goal would be to reach a same level of performance as animals, e.g. in terms of speed, cost of transport, or versatility. Much research has been done on bio-mechanical and neuro-physiological research on legged vertebrates. Hence, data is available for animal locomotion such as gait patterns, speed, cost of transport, duty factor, joint angles, torque patterns, body angles, and ground reaction force (GRF) data. While this data allows one to study a subset of locomotion characteristics, it often lacks an intuitive way to compare animals of different species, or as for us, quadruped robots. We started applying the collision angle analysis (Lee, Bertram, et al. 2011) for trot gait, based on qualitative and quantitative results from goats and dogs (taken from (ibid.)), and experimental recordings of our robot’s center of mass (COM) and GRF.