Handling multi-objective in an HTN planner for emergency decision-making

There are many multi-objective planning problems in emergency decision-making domain. HTN planners were widely used in emergency decision-making, while they have limited ability to solve optimization, especially multi-objective optimization. Aimed to handling multi-objective in HTN planners, this paper proposes a novel method based on SHOP2, which is a domain-independent state-based forward HTN planner. The method uses a weighted vector to mirror the decision maker´s performance for various objectives and an anytime search algorithm to improve the ability to find the best solution plan. Finally, a case of emergency evacuation is given to testify the effectiveness of the method.