AGPNet -- Autonomous Grading Policy Network

TL;DR

This paper introduces AGPNet, a hybrid reinforcement learning agent for autonomous dozer grading, which achieves human-level performance and generalizes well to real-world scenarios.

Contribution

It formalizes autonomous grading as a Markov Decision Process and develops a hybrid learning approach outperforming existing methods.

Findings

AGPNet reaches human-level performance.

The agent outperforms current state-of-the-art methods.

AGPNet generalizes to unseen real-world scenarios.

Abstract

In this work, we establish heuristics and learning strategies for the autonomous control of a dozer grading an uneven area studded with sand piles. We formalize the problem as a Markov Decision Process, design a simulation which demonstrates agent-environment interactions and finally compare our simulator to a real dozer prototype. We use methods from reinforcement learning, behavior cloning and contrastive learning to train a hybrid policy. Our trained agent, AGPNet, reaches human-level performance and outperforms current state-of-the-art machine learning methods for the autonomous grading task. In addition, our agent is capable of generalizing from random scenarios to unseen real world problems.