Efficient Representations of Object Geometry for Reinforcement Learning of Interactive Grasping Policies

TL;DR

This paper introduces a reinforcement learning framework for interactive robotic grasping that leverages explicit and implicit object geometry representations, enabling generalization to novel objects and complex scenarios like cluttered environments.

Contribution

It proposes a novel RL approach using signed distances for object geometry, improving grasping performance and generalization over prior model-based methods.

Findings

Successfully learned grasping policies for diverse real-world objects.

Demonstrated grasping in cluttered environments with emergent behaviors.

Showed that implicit geometric representations guide effective policy learning.

Abstract

Grasping objects of different shapes and sizes - a foundational, effortless skill for humans - remains a challenging task in robotics. Although model-based approaches can predict stable grasp configurations for known object models, they struggle to generalize to novel objects and often operate in a non-interactive open-loop manner. In this work, we present a reinforcement learning framework that learns the interactive grasping of various geometrically distinct real-world objects by continuously controlling an anthropomorphic robotic hand. We explore several explicit representations of object geometry as input to the policy. Moreover, we propose to inform the policy implicitly through signed distances and show that this is naturally suited to guide the search through a shaped reward component. Finally, we demonstrate that the proposed framework is able to learn even in more challenging conditions, such as targeted grasping from a cluttered bin. Necessary pre-grasping behaviors such as object reorientation and utilization of environmental constraints emerge in this case. Videos of learned interactive policies are available at https://maltemosbach.github. io/geometry_aware_grasping_policies.