A New Fleet Management Approach Applied to Autonomous Mining Vehicles Using Q-Learning

Abstract

In modern mining practices, a new goal for the industry is a fully autonomous mine. Most original equipment manufacturers (OEMs) are working towards automating their equipment or parts of the machines' functionality. The cooperation of an autonomous fleet would increase production, worker safety, and overall efficiency. A centralized computer system monitoring each piece of equipment’s individual data stream and system overall can ultimately replace today’s dispatch models. The new system can automatically dispatch all haul trucks based on measured conditions around the mine and adapts to new changes as they happen. Linear programming based dispatch systems are predetermined and require substantial time to reach optimization. These older models are rigid and inflexible; therefore, they do not adapt to the ever-changing mine environment. The proposed system can dynamically update the dispatching plan based on maintenance needs, payload material grade, vehicle queues, and traffic patterns. The centralized computer system is controlled by an artificial intelligence (AI) algorithm, called Q-Learning, that can optimize fleet dispatching in real time. The AI manages the truck dispatch by altering vehicles’ paths, changing loading and drop-off points, and managing maintenance needs. This algorithm must be trained for each installation and mine to suit individual needs and different fleet configurations. However, as fleet sizes change, knowledge from agents can be easily shared with new agents, and unnecessary trucks can be removed without affecting the entire system. A scaled-down proof of concept was created with autonomous robots to showcase the AI and fleet behaviours for specific scenarios to demonstrate the performance of the dynamically updating Q-learning based approach to dispatching.