ROB-GY 6203 Robot Perception

3 Credits

Smart automation systems (e.g., driverless cars, domestic/warehouse mobile robots, intelligent transportation systems, robotic construction machines, etc.) need to understand both their own poses and the surroundings to fulfil their tasks safely, accurately, and efficiently. This requires an intelligent extraction of both geometric and semantic information from sensory input (mainly visual sensors such as cameras/LIDAR). This course aims to combine the established theories of geometric vision and the recent progress in pattern recognition in the context of robotic/intelligent systems. Students will study and practice the basic theories of computer vision and machine learning through relevant applications. For example, pose estimation of a robotic agent from onboard cameras, 3D reconstruction for map creation, object detection/segmentation for obstacle avoidance, tracking for target following, place recognition from images when GPS is unreliable, and so on.

Prerequisite(s): Graduate Standing
Weekly Lecture Hours: 3