Intelligent Routing Based on Experience
This project uses previous route traversals to augment conventional route planning for military applications. To accomplish this, an intelligent route planner is being developed that learns from knowledge derived from previous traversals with an existing model-based route planner. Model-based planners have a hard time dealing with dynamic information such as weather, traffic patterns, construction or battle damage, and the location of recent attacks or threats. Some terrain features, for example fences, or paths through mountainous areas, may not appear in a model-based system at all, at least not until they are reported from the field, because they are not visible in remote sensing information such as aerial imagery. Our research exploits the insight that these domain features will be necessarily reflected in executed routes.