Object Detection for Autonomous Dozers

TL;DR

This paper presents the development and benchmarking of object detection models for autonomous dozers, aiming to improve safety and efficiency in construction site tasks through data collection, analysis, and model evaluation.

Contribution

It introduces a new autonomous dozer platform, collects construction site data, and benchmarks object detection models with various training strategies.

Findings

Object detection models achieve high accuracy on construction site data

Data analysis reveals key distribution characteristics of construction environments

Benchmarking identifies optimal training strategies for robust detection

Abstract

We introduce a new type of autonomous vehicle - an autonomous dozer that is expected to complete construction site tasks in an efficient, robust, and safe manner. To better handle the path planning for the dozer and ensure construction site safety, object detection plays one of the most critical components among perception tasks. In this work, we first collect the construction site data by driving around our dozers. Then we analyze the data thoroughly to understand its distribution. Finally, two well-known object detection models are trained, and their performances are benchmarked with a wide range of training strategies and hyperparameters.