As Artificial Intelligence progresses, Autonomous legged robots are becoming ever more desirable for operating in unstructured environments. Irregular ground, stairs, constantly changing surroundings, hazardous locations, outdoor conditions - all of these require dynamic and adaptive locomotion frameworks that enable robots to react quickly and maneuver flexibly, surpassing obstacles and replanning their motion on-the-fly. For this project, I implemented a generalized locomotion algorithm for a suit of floating-base legged robots. The algorithm allowed them to autonomously control their motion in three different levels: moving their joints coherently to achieve crawling, walking and running gaits; planning a spatial trajectory to achieve dynamically changing locations; and adjusting their routes to avoid colliding with obstacles on the way.

Developed at ETH Zürich - Switzerland, alongside one more Masters student

свернуть