Target Detection and Position Likelihood using an Aerial Image Sensor

Sensor-based control is an emerging challenge in UAV applications. It is essential in a sensing task to account for sensor measurement errors when computing a target position estimate. Source of measurement error includes those in vehicle position and orientation measurements as well as algorithm failures such as missed detections or false detections. Incorporating such errors in aerial sensors is non-trival because of the camera’s perspective geometry. We present a method to incorporate such errors into target position estimates and a calibration methodology to measure the error distributions.

Bayesian Reasoning to Estimate Target Position: Two Cases

Following errors are considered:

Position Likelihood Use when target was detected at U in the image coordinates P(X | detected, U, measurements) =>     P(U|X, measurement) If target was at X, would it be detected at U?

Detection Likelihood Use when target was not detected in the image P(X | ~detected, measurement) =>     P(~detected | X, measurement) If target was at X, would it not be detected?

Result and Applications

[Result video clip]

Example application: two-plane target search

Windows Media Player may not play the video files properly. Please use RealPlayer instead.

Publication

[Home]