ShapeMap 3-D: Efficient shape mapping through dense touch and vision

TL;DR

This paper introduces ShapeMap 3-D, a method combining tactile and visual data to efficiently create detailed 3-D object models for robotic manipulation, even in occluded or unstructured environments.

Contribution

It presents a novel incremental shape mapping approach using dense tactile sensing and deep learning, integrated with a Gaussian process-based spatial inference framework.

Findings

Effective 3-D reconstructions in simulation and real-world tests

Improved shape accuracy with combined tactile and visual data

Real-time incremental mapping capability

Abstract

Knowledge of 3-D object shape is of great importance to robot manipulation tasks, but may not be readily available in unstructured environments. While vision is often occluded during robot-object interaction, high-resolution tactile sensors can give a dense local perspective of the object. However, tactile sensors have limited sensing area and the shape representation must faithfully approximate non-contact areas. In addition, a key challenge is efficiently incorporating these dense tactile measurements into a 3-D mapping framework. In this work, we propose an incremental shape mapping method using a GelSight tactile sensor and a depth camera. Local shape is recovered from tactile images via a learned model trained in simulation. Through efficient inference on a spatial factor graph informed by a Gaussian process, we build an implicit surface representation of the object. We demonstrate visuo-tactile mapping in both simulated and real-world experiments, to incrementally build 3-D reconstructions of household objects.