CodeSLAM - Learning a Compact, Optimisable Representation for Dense Visual SLAM

TL;DR

CodeSLAM introduces a compact, learnable scene geometry representation conditioned on image data, enabling efficient dense SLAM with joint optimization of pose and local geometry codes for improved global consistency.

Contribution

The paper presents a novel, compact, learnable dense scene representation conditioned on image data, optimized jointly with pose for monocular SLAM.

Findings

Enables efficient joint optimization of pose and scene codes

Produces dense depth maps from minimal parameters

Demonstrates improved monocular SLAM performance

Abstract

The representation of geometry in real-time 3D perception systems continues to be a critical research issue. Dense maps capture complete surface shape and can be augmented with semantic labels, but their high dimensionality makes them computationally costly to store and process, and unsuitable for rigorous probabilistic inference. Sparse feature-based representations avoid these problems, but capture only partial scene information and are mainly useful for localisation only. We present a new compact but dense representation of scene geometry which is conditioned on the intensity data from a single image and generated from a code consisting of a small number of parameters. We are inspired by work both on learned depth from images, and auto-encoders. Our approach is suitable for use in a keyframe-based monocular dense SLAM system: While each keyframe with a code can produce a depth map, the code can be optimised efficiently jointly with pose variables and together with the codes of overlapping keyframes to attain global consistency. Conditioning the depth map on the image allows the code to only represent aspects of the local geometry which cannot directly be predicted from the image. We explain how to learn our code representation, and demonstrate its advantageous properties in monocular SLAM.