nextuppreviouscontents
Next:Extraction of Features from Up:Computer Vision Previous:Virtualizing Reality

Synthetic Vision System for Search and Rescue Helicopters

Certain tasks, such as flying a helicopter in proximity to the ground, are difficult to automate due to the complex interaction between the operator and the machine. For example, a pilot needs a clear view of the surrounding terrain in order to avoid obstacles, plan escape routes or to maintain the aircraft in a precise hover relative to the ground. Low to zero visibility conditions make it difficult if not impossible to fly, and are particularly problematic for search and rescue missions which often take place during less than ideal weather conditions.

One approach to solving this problem is to provide the pilot with a synthetic view of the ground by means of a helmet mounted display (immersive virtual reality). In the CAE SVS, position and heading provided by the aircraft navigation system are used to index a database of digital elevation maps from which a synthetic view of the ground is generated. This view is augmented by a live feed from an infrared scanner mounted on an external gimbal, which servos according to the pilot's direction of gaze. As of this writing we have recently completed the evaluation of different methods of fusing live + synthetic data in order to maximize the information content of the synthetic view with respect to flight control and mission task requirements. These will shortly be incorporated into the flight test platform.

The next phase of this research will involve incorporation of radar or lidar sensors, with the aim of automatically detecting potential obstacles and augmenting the display with suitable markers. Much of this will be based on our prior research in active vision and gaze planning in order to develop suitable algorithms to steer the sensors and extract the relevant information within time constraints.

P. Simard, F.P. Ferrie, N.K. Link (CAE), R. Kruk (CAE)


nextuppreviouscontents
Next:Extraction of Features from Up:Computer Vision Previous:Virtualizing Reality
Annual Report

Mon Jun 26 21:22:20 GMT 2000