Abstract
Reaching-and-grasping is a fundamental skill for robotic manipulation, but existing methods usually train models on a specific grip per and cannot be reused on another gripper. In this paper, we propose a novel method that can learn a unified policy model that can be easily transferred to different dexterous grippers. Our method consists of two stages: a gripper-agnostic policy model that predicts the displacements of pre-defined key points on the gripper, and a gripper-specific adaptation model that translates these displacements into adjustments for control ling the grippers' joints. The gripper state and interactions with objects are captured at the finger level using robust geometric representations, integrated with a transformer-based network to address variations in gripper morphology and geometry. In the experiments, we evaluate our method on several dexterous grippers and diverse objects, and the result shows that our method significantly outperforms the baseline methods. Pioneering the transfer of grasp policies across dexterous grippers, our method effectively demonstrates its potential for learning generalizable and transferable manipulation skills for various robotic hands.
Method Overview
Our method builds on a two-stage hierarchical framework that separates the prediction of unified grasps from the control of specific grippers. Given the context of the scene and the configuration of the gripper, our method initially extracts gripper- agnostic features. These features are uniformly sent to the policy model to predict gripper-agnostic point displacements, which are forwarded to the adaptation models of various grippers for precise gripper control.
Visual Results
We found that the policy trained through our proposed pipeline can achieve generalization across different dexterous hands while maintaining generalization to different objects.