Equivariant Volumetric Grasping

TL;DR

This paper introduces a rotation-equivariant volumetric grasp model using tri-plane features, improving sampling efficiency and outperforming non-equivariant methods in real-time robotic grasping tasks.

Contribution

It presents a novel tri-plane feature design with equivariance properties and adapts existing grasp planners to leverage these features for enhanced performance.

Findings

Significant improvement in sampling efficiency.

Higher grasping performance compared to non-equivariant models.

Reduced computational and memory costs.

Abstract

We propose a new volumetric grasp model that is equivariant to rotations around the vertical axis, leading to a significant improvement in sampling efficiency. Our model employs a tri-plane volumetric feature representation -- i.e., the projection of 3D features onto three canonical planes. We introduce a novel tri-plane feature design in which features on the horizontal plane are \emph{equivariant} to $90^\circ$ rotations, while the \emph{sum} of features from the other two planes remains \emph{invariant} to reflections induced by the same transformations. We further develop equivariant adaptations of two state-of-the-art volumetric grasp planners, GIGA and IGD. Specifically, we derive a new equivariant formulation of IGD's deformable attention mechanism and propose an equivariant generative model of grasp orientations based on flow matching. We provide a detailed analytical justification of the proposed equivariance properties and validate our approach through extensive simulated and real-world experiments. Our results demonstrate that the proposed projection-based design reduces both computational and memory costs. Moreover, the equivariant grasp models built on top of our tri-plane features consistently outperform their non-equivariant counterparts, achieving higher performance within a real-time cost constraint. Video and code can be viewed in: https://mousecpn.github.io/evg-page/