Signed Distance Fields: A Natural Representation for Both Mapping and Planning

Abstract

How to represent a map of the environment is a key question of robotics.In this paper, we focus on suggesting a representation well-suited for online map building from vision-based data and online planning in 3D.We propose to combine a commonly-used representation in computer graphics and surface reconstruction, projective Truncated Signed Distance Field (TSDF), with a representation frequently used for collision checking and collision costs in planning, Euclidean Signed Distance Field (ESDF), and validate this combined approach in simulation.We argue that this type of map is better-suited for robotic applications than existing representations.