Although joysticks on motorized wheelchairs have improved the lives of so many, patients with Parkinson's, stroke, limb injury, or vision problems need alternate solutions. Further, navigating wheelchairs through cluttered environments without colliding into objects or people can be a challenging task. Due to these reasons, many patients are reliant on a caretaker for daily tasks. To aid persons with disabilities, the Machine Intelligence Laboratory Personal Electronic Transport (Milpet), provides a solution. Milpet is an effective access wheelchair with speech recognition capabilities. Commands such as ``Milpet, take me to room 237’’ or ``Milpet, move forward’’ can be given. As Milpet executes the patient’s commands, it will calculate the optimal route, avoid obstacles, and recalculate a path if necessary.
This thesis describes the development of modular obstacle avoidance and path planning algorithms for indoor agents. Due to the modularity of the system, the navigation system is expandable for different robots. The obstacle avoidance system is configurable to exhibit various behaviors. According to need, the agent can be influenced by a path or the environment, exhibit wall following or hallway centering, or just wander in free space while avoiding obstacles. This navigation system has been tested under various conditions to demonstrate the robustness of the obstacle and path planning modules. A measurement of obstacle proximity and destination proximity have been introduced for showing the practicality of the navigation system. The capabilities introduced to Milpet are a big step in giving the independence and privacy back to so many who are reliant on care givers or loved ones.
Computer Engineering (MS)
Department, Program, or Center
Computer Engineering (KGCOE)
Kangutkar, Rasika M., "Obstacle Avoidance and Path Planning for Smart Indoor Agents" (2017). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus