We present a novel vision-based technique forudnavigating an Unmanned Aerial Vehicle (UAV) through urbanudcanyons. Our technique relies on both optic flow and stereoudvision information. We show that the combination of stereo and optic-flow (stereo-flow) is more effective at navigating urban canyons than either technique alone. Optic flow from a pair of sideways-looking cameras is used to stay centered in audcanyon and initiate turns at junctions, while stereo vision fromuda forward-facing stereo head is used to avoid obstacles to theudfront. The technique was tested in full on an autonomous tractorudat CSIRO and in part on the USC autonomous helicopter.udExperimental results are presented from these two roboticudplatforms operating in outdoor environments. We show that theudautonomous tractor can navigate urban canyons using stereoflow,udand that the autonomous helicopter can turn away fromudobstacles to the side using optic flow. In addition, preliminaryudresults show that a single pair of forward-facing fisheye camerasudcan be used for both stereo and optic flow. The center portionsudof the fisheye images are used for stereo, while flow is measuredudin the periphery of the images.
展开▼