The Design and Implementation of a Scheduler and Route Planner for Wheelchair Users

Currently, Global Positioning Systems and other route-finding systems work well for wheelchair users outdoors. Indoor information is usually not provided. However, it is more important for wheelchair users to know the indoor maps of routes accessible to them. Where the closest elevator is, which door is accessible to them, and how long it takes to get to a destination are major route-finding problems for wheelchair users. Nevertheless, there is no existing system that shows up-to-date and detailed information on route accessibility. A system that works for the independent user would provide more accurate help for wheelchair users. For example, a strong young man in a wheelchair will have less difficulty getting through moderate ramps and rough roads than the elderly who tend to prefer smooth routes. Therefore, we are motivated to create a system that works according to the client’s ability, so that it can figure out which path is best for the client under certain constraints. There are several existing systems that allow the user to easily control their wheelchairs. Some of these systems use automatic or semi-automatic robotic wheelchairs, but they are limited to very small local area movement. Such systems focus on hardware components of an auto-wheelchair. Our system, which provides help in pathfinding, can cooperate with these hardware-equipped wheelchairs. The success of related work on daily activity reminder systems has motivated us to build a simple scheduler that works in conjunction with the route planner. As well as designing and creating the new system, we have analyzed and implemented a new efficient pathfinding algorithm, which is the major contribution of this thesis. The algorithm deploys the hierarchical structure of real-life maps. Assuming that the distance within an abstract high-level node is significantly shorter than along high-level edges, the algorithm can prune away irrelevant paths. Runtime analysis and experimental results show that this algorithm is efficient in numerous scenarios.