CS 481B - Special Topics

Credits: 3

Robot Perception Perception is the ability to perceive, understand, and reason about the surrounding environment. Therefore, a solid perception ability is necessary to enable robot autonomy in the real-world, e.g., self-driving cars, aerial drones, autonomous boats. This course offers a holistic understanding of the fundamentals as well as the latest trends in robot perception. Initial topics include fundamentals of robotics, sensing techniques, 3D representations, state estimation, sensor fusion, SLAM, and learning algorithms. Then, the class will discuss open problems in robot perception, including object detection and tracking, semantic scene understanding, pose estimation, perception in dynamic environments, open-world perception, perception for manipulation and navigation, sim2real, and uncertainty modeling. This course is heavily discussion and team project oriented. There will be a mix of lectures, discussions, and student presentations. Students are responsible for paper readings, in-class discussion, and completing a semester-long project. Prerequisite: CS 375, MATH 304. Recommended: CS 350 ? for knowledge of Linux-based operating systems, A background in Computer Vision, Machine Learning, and Robotics is a plus, but not required, and the basics will be introduced in context. Familiarity with the Robot Operating Systems (ROS) will be beneficial. Expected to be offered once a year.