Mobile robot tele-operation support using gaze intention estimator with saliency map and outlier test
Reo Arita, Satoshi V. Suzuki · 2020
Narrow angle of view of the camera mounted on the mobile remote-controlled robots makes the operation difficult. In order to solve this problem, a panning support system utilizing the followings is presented in this paper: estimation of operator's attention movement and estimation of gaze point candidate. In this system, the gaze movement is detected using a Kullback-Leibler outlier test, and the gaze points are estimated utilizing a saliency map. This paper introduces the configuration of this support system and the outlier test method using Kullback-Leibler, and the generation of the saliency maps. The primely measurement / analysis using a mobile robot operation experiment in virtual space was performed to confirm the presented algorithms and methods.