Object Detection and Classification in Occupancy Grid Maps using Deep Convolutional Networks

TL;DR

This paper presents a deep learning approach for object detection and classification in occupancy grid maps derived from sensor data, achieving state-of-the-art accuracy for automated vehicle perception.

Contribution

It introduces a novel method using deep convolutional networks on multi-layer grid maps for robust object detection in autonomous driving environments.

Findings

Achieves state-of-the-art detection accuracy on KITTI benchmark.

Demonstrates robustness of grid map-based detection methods.

Provides insights through extensive ablation studies.

Abstract

A detailed environment perception is a crucial component of automated vehicles. However, to deal with the amount of perceived information, we also require segmentation strategies. Based on a grid map environment representation, well-suited for sensor fusion, free-space estimation and machine learning, we detect and classify objects using deep convolutional neural networks. As input for our networks we use a multi-layer grid map efficiently encoding 3D range sensor information. The inference output consists of a list of rotated bounding boxes with associated semantic classes. We conduct extensive ablation studies, highlight important design considerations when using grid maps and evaluate our models on the KITTI Bird's Eye View benchmark. Qualitative and quantitative benchmark results show that we achieve robust detection and state of the art accuracy solely using top-view grid maps from range sensor data.