AI for rover terrain traversability
Video courtesy of research partner Mission Control Space Services Inc.
Machine learning and AI provide opportunities to gain novel insights into terrain properties from existing rover sensors, including cameras and Inertial Measurement Units (IMUs). Our research in this area includes vision-based novelty detection using Autoencoders, IMU-vibration-based slip detection, crater-based navigation, and analysis of specific image features that sometimes confuse neural networks regarding the type of terrain (e.g. sand vs rocks) they are encountering. Our research is tailored to planetary rovers and their significant sensing and computing limitations, seeking balance in both performance and practicality.
Machine learning and AI provide opportunities to gain novel insights into terrain properties from existing rover sensors, including cameras and Inertial Measurement Units (IMUs). Our research in this area includes vision-based novelty detection using Autoencoders, IMU-vibration-based slip detection, crater-based navigation, and analysis of specific image features that sometimes confuse neural networks regarding the type of terrain (e.g. sand vs rocks) they are encountering. Our research is tailored to planetary rovers and their significant sensing and computing limitations, seeking balance in both performance and practicality.
Concordia University's Aerospace Robotics Lab
Research interests include: Space robotics, Planetary rovers, Robot mobility, Vehicle-terrain interactions, Advanced 3D printing techniques, Robotics excavation & construction, Reduced gravity experimentation, Computer vision and machine learning for robotics applications.