Experimental analysis of Radar/Optical track-to-track fusion for non-cooperative Sense and Avoid

In the framework of non-cooperative Sense and Avoid solutions, major attention is reserved to the design of sensing strategies which can enable fast and reliable identification of possible near collision threats, by exploiting passive or active exteroceptive sensors. To overcome the limits of standalone technological architectures and provide more accurate tracking solutions, which can be used to improve conflict detection and thus better support avoidance strategies, data fusion approaches can be considered. Hence, this work proposes a radar/visual fusion method based on the track-to-track fusion approach. The strategy is tested on data gathered during ground-to-air experimental flight tests involving a small UAV commanded to fly near collision approach geometries with respect to a multi-sensor setup placed on the ground. Results collected analyzing three different encounters show that the fusion solution allows retrieving meter and sub-degree level accuracies in the intruder range and bearing estimation, respectively, while ensuring declaration ranges of about 500 meters.