Online and Certifiably Correct Visual Odometry and Mapping

TL;DR

This paper introduces two algorithms for certified visual odometry and mapping using RGBD cameras, providing provable safety guarantees and demonstrating real-time performance in hardware experiments.

Contribution

It presents the first algorithms that offer certifiable error bounds for visual odometry and obstacle mapping in safety-critical robotic applications.

Findings

Algorithms run online at 30FPS

Demonstrated in hardware experiments

Compared favorably to state-of-the-art methods

Abstract

This paper proposes two new algorithms for certified perception in safety-critical robotic applications. The first is a Certified Visual Odometry algorithm, which uses a RGBD camera with bounded sensor noise to construct a visual odometry estimate with provable error bounds. The second is a Certified Mapping algorithm which, using the same RGBD images, constructs a Signed Distance Field of the obstacle environment, always safely underestimating the distance to the nearest obstacle. This is required to avoid errors due to VO drift. The algorithms are demonstrated in hardware experiments, where we demonstrate both running online at 30FPS. The methods are also compared to state-of-the-art techniques for odometry and mapping.