Point cloud obstacle detection with the map filtration

TL;DR

This paper introduces a point cloud obstacle detection pipeline using map filtration, enabling real-time obstacle detection on limited hardware, validated through real robot testing and dataset evaluation.

Contribution

The paper presents a novel obstacle detection pipeline based on map filtration that operates efficiently on low-power devices, demonstrated through real robot experiments and dataset analysis.

Findings

Effective obstacle detection on limited hardware

Comparable performance to 3D object detectors

Validated with real robot and dataset evaluations

Abstract

Obstacle detection is one of the basic tasks of a robot movement in an unknown environment. The use of a LiDAR (Light Detection And Ranging) sensor allows one to obtain a point cloud in the vicinity of the sensor. After processing this data, obstacles can be found and recorded on a map. For this task, I present a pipeline capable of detecting obstacles even on a computationally limited device. The pipeline was also tested on a real robot and qualitatively evaluated on a dataset, which was collected in Brno University of Technology lab. Time consumption was recorded and compared with 3D object detectors.