| Summary: | This paper focuses on an approach to dynamic planning, for an emergency ambient such as team rescue in indoors fire. First a graph is generated as the simulation runs, creating an effect similar to means-end analysis as each fire trying to reach the firefighter. This graph is updated in real time, improving the solution performance and reacting to new fires. The firefighter creates a plan based on this graph, using shortest weighed paths algorithms, these weights are updated dynamically, they do not only contain the distance but they also contain the importance to reach that node, so a important node to reach costs less for a firefighter to get there. All this together allows real time solutions to be generated, and self improving solutions to be made in the plan. This algorithm is to be integrated on a framework that simulates physics and collisions, and using a navigation mesh and agent perceptions to aid in calculation of a 3D shortest path.
|
|---|