MineSim: A scenario-based simulation test system and benchmark for autonomous trucks in open-pit mines

Area of Science:

• Robotics and Autonomous Systems
• Simulation and Modeling
• Mining Engineering

Background:

• Autonomous driving simulators are crucial for testing but often lack coverage of unstructured, non-urban environments.
• Open-pit mining presents unique challenges for autonomous vehicles, including irregular roads and specific operational demands.

Purpose of the Study:

• Introduce MineSim, an open-source, scenario-based simulation test system tailored for autonomous trucks in open-pit mines.
• Provide a platform for validating planning algorithms and evaluating system performance in complex mining scenarios.

Main Methods:

• Developed automated scenario parsing, state update models, agent policies, and metric evaluation tools.
• Incorporated real-world traffic scenarios from two open-pit mines.
• Created scenario libraries and benchmarks for obstacle avoidance tasks.

Main Results:

• MineSim effectively simulates unstructured road environments with unique challenges like irregular intersections and lack of lane markings.
• The system captures the specific dynamics of heavy autonomous mining trucks, including response lags.
• Benchmarks for static and dynamic obstacle avoidance facilitate algorithm research.

Conclusions:

• MineSim provides a vital resource for advancing autonomous driving in challenging, unstructured, non-urban settings.
• The simulation system enables reproducible testing and data sharing for the mining industry.
• Facilitates the development and validation of safer and more efficient autonomous mining operations.