MoGaze: A Dataset of Full-Body Motions that Includes Workspace Geometry and Eye-Gaze

As robots become more present in open human environments, it will crucial for robotic systems to understand and predict motion. Such capabilities depend heavily on the quality availability of motion capture data. However, existing datasets full-body rarely include 1) long sequences manipulation tasks, 2) 3D model workspace geometry, 3) eye-gaze, which are all important when a robot needs movements humans close proximity. Hence, this letter, we novel dataset everyday includes above. The data was captured using traditional system based reflective markers. We additionally eye-gaze wearable pupil-tracking device. show experiments, can be used design evaluation prediction algorithms. Furthermore, our experiments as powerful predictor intent. 180 min with 1627 pick place actions being performed. It is available at https://humans-to-robots-motion.github.io/mogaze/MoGaze, Dataset planned extended collaborative tasks two near future.