Learning Motion Feasibility from Point Clouds in Cluttered Environments

2026-06-25

AI Summary

This work addresses the inefficiency of traditional sampling-based motion planners in cluttered environments, where numerous infeasible attempts arise due to limitations of existing feasibility classifiers—often restricted to low-dimensional configuration spaces and simplified geometric assumptions. To overcome these constraints, the authors introduce the first large-scale grasp feasibility benchmark comprising 2.7 million labeled samples, enabling end-to-end learning of high-dimensional motion feasibility for 7-DoF robotic arms directly from raw RGB-D point clouds. Through a systematic comparison of MLP, voxel-based CNN, and point cloud Transformer architectures, the proposed model, GRASPFC-PTX, achieves an AUROC of 0.996 on novel objects and significantly outperforms conventional sampling-based planners in prediction speed, thereby eliminating reliance on simplified environmental models and low-dimensional configuration spaces.

Abstract

Motion feasibility prediction plays a central role in robotics, particularly in task and motion planning and manipulation. A major bottleneck for this problem in cluttered environments is that infeasible planning attempts by Sampling-based motion planners (SBMPs) can incur substantial computational cost. Also existing approaches for infeasibility certification are limited to low-dimensional configuration spaces and often assume simplified geometric environments represented by primitive objects with known parameters. We study the complementary problem of learning motion feasibility prediction directly from raw RGB-D observations for a 7-DOF manipulator operating in realistic cluttered scenes. We introduce the first large-scale benchmark for this setting, comprising 2.7M grasp feasibility labels over 88 scanned objects and 190 cluttered tabletop scenes. We benchmark three representative classifier families spanning MLP- based, volumetric-CNN, and point-cloud-based Transformer architectures under matched training conditions. Our best model, GRASPFC-PTX (a point-cloud transformer), achieves an AUROC of 0.996 on Novel objects while providing predictions significantly faster than SBMPs.

Problem

• motion feasibility
• cluttered environments
• point clouds
• robot manipulation
• task and motion planning

Innovation

• motion feasibility prediction
• point cloud transformer
• cluttered environments
• grasp feasibility
• large-scale benchmark