Real time motion recovery using a hemispherical sensor

This paper describes work in a new project based on a collaboration between experts in low vision and a computer vision research group. The focus of the project is to develop assistive devices for individuals with severe and profound vision impairment resulting from diseases such as Age-related Macular Degeneration and Retinitis Pigmentosa. We describe focus groups that are being conducted to understand such needs. To assist with tasks such as navigation and obstacle avoidance for an individual who is walking, knowledge of self-motion is essential. In this context we present a new implementation of a wide angle camera visual motion recovery algorithm suitable for use on a low cost, low power, light-weight wearable sensing device. For wearable sensing, camera paths are far more erratic than for ground based vehicles such as wheeled robots or cars. Also, weight from computing, cameras and batteries is a major issue.