Agent-based transportation planning compared with scheduling heuristics

Here we consider the problem of dynamically assigning vehicles to transportation orders that have di¤erent time windows and should be handled in real time. We introduce a new agent-based system for the planning and scheduling of these transportation networks. Intelligent vehicle agents schedule their own routes. They interact with job agents, who strive for minimum transportation costs, using a Vickrey auction for each incoming order. We use simulation to compare the on-time delivery percentage and the vehicle utilization of an agent-based planning system to a traditional system based on OR heuristics (look-ahead rules, serial scheduling). Numerical experiments show that a properly designed multi-agent system may perform as good as or even better than traditional methods.