Perceptual Anchoring for Gaze-Tracking Wearables and Robot-Mounted Sensors

For persons with motor disabilities, robots havea great potential to improve the quality of daily interactions.A fully-fledged solution requires addressing many differentchallenges. Gaze-tracking wearables provides a new inputmodality for such solutions. In this paper we address thechallenge of building and maintaining a world model usingrobot and gaze-tracking sensing data. We construct a worldmodeling pipeline consisting of a data association step and ananchoring step. In the data association step we compare fourdifferent methods. The results show that three possible dataassociation methods outperform a geometric baseline. However,the results have potential for improvement, and thus futurework to enhance the performance of the pipeline is needed.To guide future work, we have investigated potential factorswhich affect the performance of both the data association andthe anchoring process, and found that the positions of detectedobjects have a significant impact.