We are developing autonomous technology for off-road driving in wilderness environments Key developments include perception, planning and control capabilities. This is a joint development with Yamaha Motor Corporation and the CMU Field Robotics Center.
- Pose Estimation: Fuse multiple sensors (camera, LIDAR & IMU) with GPS to enhance accurate local mapping and to cope with GPS-degraded situations.
- Classification: Use Pose Estimation to reconstruct local geometry, use convolutional neural network approach to semantically label visible objects.
- Global Planning: Use publicly available aerial imagery and digital elevation maps to plan long trips.
- Local Planning: Build mesh model of local terrain from LIDAR data, simulate vehicle travel for candidate paths to find best local path.
|The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.|
Contact Us | Update Instructions