MISCGrasp: Leveraging Multiple Integrated Scales and Contrastive Learning for Enhanced Volumetric Grasping

TL;DR

MISCGrasp introduces a novel volumetric grasping approach that combines multi-scale feature extraction, contrastive learning, and transformer-based interactions to improve robotic grasping adaptability and performance.

Contribution

It presents a new method integrating multi-scale features with contrastive learning and transformer modules for enhanced self-adaptive volumetric grasping.

Findings

Outperforms baseline methods in simulated environments.

Achieves higher success rates in real-world tabletop tasks.

Effectively balances geometric detail and structure through transformers.

Abstract

Robotic grasping faces challenges in adapting to objects with varying shapes and sizes. In this paper, we introduce MISCGrasp, a volumetric grasping method that integrates multi-scale feature extraction with contrastive feature enhancement for self-adaptive grasping. We propose a query-based interaction between high-level and low-level features through the Insight Transformer, while the Empower Transformer selectively attends to the highest-level features, which synergistically strikes a balance between focusing on fine geometric details and overall geometric structures. Furthermore, MISCGrasp utilizes multi-scale contrastive learning to exploit similarities among positive grasp samples, ensuring consistency across multi-scale features. Extensive experiments in both simulated and real-world environments demonstrate that MISCGrasp outperforms baseline and variant methods in tabletop decluttering tasks. More details are available at https://miscgrasp.github.io/.