Event
MRC Student Seminar: A Control Theoretic Approach to Monocular Computer Vision
Friday, October 14, 2022
2:00 p.m.
AVW-1146
LaToya Hale
301 405 3108
lthale@umd.edu
The Advantages of a Control Theoretic Approach to Monocular Computer Vision
Levi Burner
PhD Student
Electrical & Computer Engineering Department
Perception and Robotics Group
The University of Maryland
Abstract
It is well known that for certain computer vision tasks, such as a monocular distance estimation, some self-motion is essential. In robotics this motion is frequently due to a control policy. However, the defacto implementation pattern in robotics treats state estimation and control as separate problems. Thus, this talk will explore the advantages of a joint approach to computer vision and control, including a linear equality constraint that makes state estimation efficient and accurate, a stability invariance property that makes it easier for low-level controllers to remain stable, and an empirical improvement where 30-70% less trajectory estimation error was achieved with algorithm running 6-25 times faster than algorithms with comparable purpose. The technical discussion will be specifically tailored towards enabling the audience to adopt these discussed methods in their own robotics projects. These results are described in the following paper: https://arxiv.org/abs/2203.07530
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.
