TacLoc: Global Tactile Localization on Objects from a Registration Perspective

TL;DR

TacLoc is a novel tactile localization framework that formulates pose estimation as a point cloud registration problem, achieving accurate results without pre-trained models or rendered data, and demonstrating effectiveness on real-world objects.

Contribution

It introduces a graph-theoretic partial-to-full registration method leveraging tactile point clouds, normal-guided pruning, and a hypothesis-verification pipeline for improved tactile pose estimation.

Findings

Achieves high accuracy on YCB dataset

Demonstrates effectiveness on real-world tactile sensors

Operates without pre-trained models or rendered data

Abstract

Pose estimation is essential for robotic manipulation, particularly when visual perception is occluded during gripper-object interactions. Existing tactile-based methods generally rely on tactile simulation or pre-trained models, which limits their generalizability and efficiency. In this study, we propose TacLoc, a novel tactile localization framework that formulates the problem as a one-shot point cloud registration task. TacLoc introduces a graph-theoretic partial-to-full registration method, leveraging dense point clouds and surface normals from tactile sensing for efficient and accurate pose estimation. Without requiring rendered data or pre-trained models, TacLoc achieves improved performance through normal-guided graph pruning and a hypothesis-and-verification pipeline. TacLoc is evaluated extensively on the YCB dataset. We further demonstrate TacLoc on real-world objects across two different visual-tactile sensors.