SCS Undergraduate Thesis Topics
|Bradford Neuman||Tony Stentz||Learning-based Change Detection for Mobile Robots|
Mobile Robotics has advanced to a stage where robotic vehicles are beginning to navigate autonomously and make decisions about the traversibility of obstacles based on rich sensor data and machine learning and planning techniques. Unfortunately, these systems can still fail, especially in circumstances on which they were not trained. My goal is to create a technique which should improve robot safety and reliability by allowing robots to detect important changes in their environment. If a robot sees a given scene more than once and there is a significant change, such as a human being present or a tree falling, the robot should be able to detect the change and avoid it or alert a human. I am investigating machine learning techniques to perform change detection on a set of rich sensor data collected by a mobile robot.