Deep semantic classification for 3D LiDAR data

Robots are expected to operate autonomously in dynamic environments. Understanding the underlying dynamic characteristics of objects is a key enabler for achieving this goal. In this paper, we propose a method for pointwise semantic classification of 3D LiDAR data into three classes: non-movable, movable and dynamic. We concentrate on understanding these specific semantics because they characterize important information required for an autonomous system. To learn the distinction between movable and non-movable points in the environment, we introduce an approach based on deep neural network and for detecting the dynamic points, we estimate pointwise motion. We propose a Bayes filter framework for combining the learned semantic cues with the motion cues to infer the required semantic classification. In extensive experiments, we compare our approach with other methods on a standard benchmark dataset and report competitive results in comparison to the existing state-of-the-art. Furthermore, we show an improvement in the classification of points by combining the semantic cues retrieved from the neural network with the motion cues.

Authors

• Ayush Dewan
• Gabriel L. Oliveira
• Wolfram Burgard

Keywords

• Semantics
• Laser radar
• Three-dimensional displays
• Training
• Two dimensional displays
• Dynamics
• Neural networks
• Lidar Data
• 3D LiDAR
• 3D LiDAR Data
• Neural Network
• Deep Neural Network
• Environmental Point
• Dynamic Point
• Improvement In Classification
• Point Classification
• Motion Cues
• Standard Benchmark Datasets
• Detection Methods
• Convolutional Neural Network
• Object Detection
• 2D Images
• Precision And Recall
• Semantic Information
• Bounding Box
• Object Classification
• Semantic Segmentation
• LiDAR Scans
• Consecutive Scans
• Point Scanning
• Objective Information
• Factor Graph
• Objects In The Scene
• Dynamic Objects
• Objective Scores
• Front Camera