Learning Visual Quadrupedal Loco-Manipulation from Demonstrations
This addresses the limitation of quadruped robots in interacting with objects in human environments, offering a more efficient approach without additional arms, though it appears incremental as it builds on existing RL and BC methods.
The paper tackled the problem of enabling quadruped robots to perform manipulation tasks using only their legs, by decomposing loco-manipulation into a low-level RL controller and a high-level BC planner, and demonstrated the robot's ability to execute tasks like lifting a basket, closing a dishwasher, pressing a button, and pushing a door in simulations and real-world experiments.
Quadruped robots are progressively being integrated into human environments. Despite the growing locomotion capabilities of quadrupedal robots, their interaction with objects in realistic scenes is still limited. While additional robotic arms on quadrupedal robots enable manipulating objects, they are sometimes redundant given that a quadruped robot is essentially a mobile unit equipped with four limbs, each possessing 3 degrees of freedom (DoFs). Hence, we aim to empower a quadruped robot to execute real-world manipulation tasks using only its legs. We decompose the loco-manipulation process into a low-level reinforcement learning (RL)-based controller and a high-level Behavior Cloning (BC)-based planner. By parameterizing the manipulation trajectory, we synchronize the efforts of the upper and lower layers, thereby leveraging the advantages of both RL and BC. Our approach is validated through simulations and real-world experiments, demonstrating the robot's ability to perform tasks that demand mobility and high precision, such as lifting a basket from the ground while moving, closing a dishwasher, pressing a button, and pushing a door. Project website: https://zhengmaohe.github.io/leg-manip