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

2025-07-03

AI Summary

To address the limited adaptability of robotic grasping to objects with diverse shapes and sizes, this paper proposes a voxel-based multi-scale contrastive learning framework for grasp planning. The method employs a dual-Transformer architecture—comprising an Insight Transformer and an Empower Transformer—that enables query-driven interaction between high-level semantic and low-level geometric features, facilitating cross-scale feature fusion. It further integrates multi-scale voxel convolutions with a contrastive learning objective to jointly optimize fine-grained geometric detail perception and holistic structural modeling. This design significantly enhances feature discriminability and cross-scale consistency. Evaluated on both simulated and real-world tabletop clutter clearing tasks, the approach achieves substantially higher grasp success rates and robustness compared to state-of-the-art baselines and ablation variants, demonstrating its effectiveness and strong generalization capability.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

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/.

Problem

• Adapting robotic grasping to varying object shapes and sizes
• Balancing fine geometric details and overall structures in grasping
• Ensuring multi-scale feature consistency via contrastive learning

Innovation

• Multi-scale feature extraction for grasping
• Contrastive learning for feature enhancement
• Query-based interaction with Insight Transformer