Towards a Multi-Embodied Grasping Agent
This work addresses the need for generalist grasping across multiple gripper types, offering a more data-efficient and scalable solution for robotic manipulation.
The authors propose a data-efficient, flow-based, equivariant grasp synthesis architecture that generalizes across diverse gripper designs by exploiting kinematic models from geometry alone, achieving smoother learning and faster inference with 25,000 scenes and 20 million grasps.
Multi-embodiment grasping focuses on developing approaches that exhibit generalist behavior across diverse gripper designs. Existing methods often learn the kinematic structure of the robot implicitly and face challenges due to the difficulty of sourcing the required large-scale data. In this work, we present a data-efficient, flow-based, equivariant grasp synthesis architecture that can handle different gripper types with variable degrees of freedom and successfully exploit the underlying kinematic model, deducing all necessary information solely from the gripper and scene geometry. Unlike previous equivariant grasping methods, we translated all modules from the ground up to JAX and provide a model with batching capabilities over scenes, grippers, and grasps, resulting in smoother learning, improved performance and faster inference time. Our dataset encompasses grippers ranging from humanoid hands to parallel yaw grippers and includes 25,000 scenes and 20 million grasps.