CLAP: Closed-Loop Alignment with Pressure for Precise Suction Manipulation

2026-09-26

AI Summary

This study addresses the challenges of suction cup grasping mislocalization caused by visual occlusion in stacked scenarios and the inability of open-loop execution to perform real-time error correction. We propose a pressure-aware Vision-Language-Action (VLA) policy that, for the first time, incorporates measured vacuum signals as conditional inputs into the VLA model. By integrating offline joint-sequence reversal data augmentation, the pressure feedback disrupts the open-loop execution window and triggers closed-loop re-reasoning to promptly correct grasping poses. Experiments conducted on a Unitree Z1 robotic manipulator demonstrate that the proposed method achieves an average success rate of 96.67%, surpassing the baseline by over ten percentage points, with a positional error of only 15.92 mm. These results confirm the effectiveness of our approach in realizing high-precision, adaptive suction control.

Abstract

Stacking and palletising demand precise placement: error left in one layer is inherited by the next, and a flat pad offers no feature to funnel a wrong pose into the right one. Top-down suction suits such dense arrangements, and suction has already been brought into vision-language-action (VLA) policies. What that work does not report, however, is a policy conditioned on a measured vacuum signal, or one that uses it to abandon an action already under way. Vision does not settle the question here, because at the moment it matters the cup and the face it holds occlude each other. We present CLAP, which makes the attachment state observable through a pressure module tapped into the vacuum line. The decoded reading replaces the suction command in the policy's proprioception, is fused with the visual features, and terminates the open-loop execution window so that the policy re-infers from a fresh observation. For data, we record goal-state disassembly on the physical robot and reverse the joint-state sequence offline, without a simulation replay. Targeted phase demonstrations, 8.3% of the training frames, cover the suction transitions and the configurations an interrupted grasp leaves behind. On a real Unitree Z1, one multi-task checkpoint reaches 96.67% average success in both colour settings, 16.67 and 10.00 points above the strongest baseline, its monochromatic four-block successes averaging 15.92 mm of error. Four ablation settings fall 5.00 to 13.33 points short. We will release code and trained weights.

Problem

• suction manipulation
• precise placement
• vacuum pressure feedback
• occlusion
• vision-language-action policy

Innovation

• Closed-loop alignment
• Pressure feedback
• Suction manipulation
• Vision-language-action policy
• Demonstration reversal