Lizards...

...co-design of robotic lizards at ETH Robotic Systems Lab


The video shows two sessions of a Tree-structured Parzen Estimator-based Bayesian Optimization for single-sized legged lizards to optimize the parameters of a controller to generate "Hildebrand-like" gaits.


The project is a collaboration between Prof. Marco Hutter and Prof. Baxi Chong (my former mentor at Dan Goldman’s lab). The objective of this project is to explore and optimize the morphological, control, and environmental parameter space in lizard-inspired robots, using simulation and learning, to optimize their speed, acceleration, and energetic cost across cluttered terrains, slopes, and sand.
The ultimate objectives are understanding evolutionary trends through a paleoinspired robotics approach (e.g. independently evolved limbless body forms), and especially informing the design of multi-modal robots that can adapt their morphology and control policies in real time to traverse diverse environments.


Performance difference of multiple-sized legged lizards using the exact same non-optimized gait.