MRC Student Seminar: Dynamically Finding Optimal Observer States to Minimize Localization Error
Friday, September 30, 2022
301 405 3108
Dynamically Finding Optimal Observer States to Minimize Localization Error with State-Dependent Noise
Computing Innovation (CI) Fellow
Sometimes, an autonomous robot is ill-equipped to perceive its surrounding environment. For example, an autonomous car with cameras, LiDAR, radar, and GPS may fail to perceive its surroundings in adverse weather. Failure to sense the environment means the robot cannot determine where it is (localize) and could hurt people or damage itself or property. We are interested in helping such robots localize when they cannot sense their environment for an extended period, reducing harm to people and property. In this talk, I will present a novel technique called DyFOS, and describe a scenario where an observer (such as a drone) helps an ill-equipped robot (a target) localize more accurately. An observer uses DyFOS to find optimal viewpoints of the target to minimize state-dependent measurement noise, predict collisions with obstacles, and predict occlusions of the target. I will present preliminary results comparing DyFOS with brute-force and random and heuristic search methods.
About the Robotics Student Seminars
The Robotics Student Seminars at the University of Maryland College Park are a student-run series of talks given by current robotics students.
The purpose of these talks is to:
- Encourage interaction between Robotics students from different subfields;
- Provide an opportunity for Robotics students to be aware of and possibly get involved in the research their peers are conducting;
- Provide an opportunity for Robotics students to receive feedback on their current research;
- Provide speaking opportunities for Robotics students.
**Light refreshments will be served