Task-oriented grasping for dexterous robots using postural synergies and reinforcement learning

TL;DR

This paper presents a reinforcement learning approach for task-oriented grasping in humanoid robots, integrating human grasp preferences and postural synergies to improve context-aware manipulation.

Contribution

It introduces an end-to-end reinforcement learning framework that incorporates human grasp data and postural synergies for improved task-specific grasping.

Findings

Successfully trained an agent to grasp multiple objects with task-specific constraints

Demonstrated the effectiveness of combining human grasp data with reinforcement learning

Achieved more natural and context-aware manipulation behaviors

Abstract

In this paper, we address the problem of task-oriented grasping for humanoid robots, emphasizing the need to align with human social norms and task-specific objectives. Existing methods, employ a variety of open-loop and closed-loop approaches but lack an end-to-end solution that can grasp several objects while taking into account the downstream task's constraints. Our proposed approach employs reinforcement learning to enhance task-oriented grasping, prioritizing the post-grasp intention of the agent. We extract human grasp preferences from the ContactPose dataset, and train a hand synergy model based on the Variational Autoencoder (VAE) to imitate the participant's grasping actions. Based on this data, we train an agent able to grasp multiple objects while taking into account distinct post-grasp intentions that are task-specific. By combining data-driven insights from human grasping behavior with learning by exploration provided by reinforcement learning, we can develop humanoid robots capable of context-aware manipulation actions, facilitating collaboration in human-centered environments.