Mechanical Design, Development and Testing of Bioinspired Legged Robots for Dynamic Locomotion

Animals like ostriches perform complex locomotion tasks in the presence of large neurocontrol delays and potentially with limited central command. It is surprising that legged robots with high control frequencies and microsecond-range control delays are still unable to compete with legged animals in terms of agility, robustness, performance, and energy efficiency. These observations led us to look closely at animals to draw inspiration and improve robotic leg designs. For example, we observed that certain leg joints moved in a coordinated manner but without neural command when manually flexing specific leg joints. We further observed that birds' legs could be loaded or slack depending on foot orientation. This mechanical clutch behavior of the foot in combination with a mechanical coupling of the leg joints was a key design element for the new generation of leg design. From our observations, we developed a multi-segment leg that behaves like a spring-loaded leg during the stance phase and disengages its elasticity at the transition into the swing phase. With this functionality, the robot leg performs its locomotion tasks efficiently with minimal central control input, unlike previous robotic legs. In this talk, besides explaining the leg mechanism and its functionality, I will present robots and testbeds developed with this leg to explain some locomotion hypotheses that are impossible to test in the animal.