Bayesian Spatial Kernel Smoothing for Scalable Dense Semantic Mapping

TL;DR

This paper introduces a Bayesian spatial kernel smoothing method for creating scalable, dense 3D semantic maps from noisy point clouds, improving smoothness and local correlation exploitation over traditional occupancy mapping.

Contribution

It generalizes Bayesian kernel inference from binary occupancy to multi-class semantic mapping, providing a unified probabilistic model that enhances map smoothness and performance.

Findings

Outperforms current baseline methods in multiple datasets

Runs at about 2 Hz on a laptop CPU

Provides qualitative results on robot-collected data

Abstract

This paper develops a Bayesian continuous 3D semantic occupancy map from noisy point clouds by generalizing the Bayesian kernel inference model for building occupancy maps, a binary problem, to semantic maps, a multi-class problem. The proposed method provides a unified probabilistic model for both occupancy and semantic probabilities and nicely reverts to the original occupancy mapping framework when only one occupied class exists in obtained measurements. The Bayesian spatial kernel inference relaxes the independent grid assumption and brings smoothness and continuity to the map inference, enabling to exploit local correlations present in the environment and increasing the performance. The accompanying software uses multi-threading and vectorization, and runs at about 2 Hz on a laptop CPU. Evaluations using multiple sequences of stereo camera and LiDAR datasets show that the proposed method consistently outperforms current baselines. We also present a qualitative evaluation using data collected with a bipedal robot platform on the University of Michigan - North Campus.