The objective of this project is learning high-level manipulation planning skills from humans and transfer these skills to robot planners. We used virtual reality to generate data from human participants whilst they reached for objects on a cluttered table top. From this, we devised a qualitative representation of the task space to abstract human decisions, irrespective of the number of objects in the way. Based on this representation, human demonstrations were segmented and used to train decision classifiers. Using these classifiers, our planner produced a list of waypoints in the task space. These waypoints provide a high-level plan, which can be transferred to any arbitrary robot model. The VR dataset is released here.

The data are kinematic measures of motor control asymmetries in 69 adults and 140 children (aged 5-6 years) from Britain and Kuwait. Data are provided on age (adult/child); nationality (British/Kuwaiti), gender (male/female); handedness (right-handed/left-handed); and task performance (tracing error score (mm) on a tracing task, where participants completed the task both in the left-to-right direction, and the right-to-left direction).

This dataset is supplied to accompany the paper entitled "Predicting the duration of reach-to-grasp movements to objects with asymmetric contact surfaces". It contains data from a series of experiments. In Experiment 1a, participants reached-to-lift wooden blocks off a table top, with the contact locations for the thumb and index finger varying in surface size. In Experiment 1b participants reached-to-grasp the wooden blocks. Experiment 2 tested whether our findings generalised to reach-to-grasp behaviour with conical frusta grasped in a different plane (thumb on top, finger on bottom) mounted off the table top. The thumb surface was the visible surface in experiments 1 and 2 so Experiment 3 explored grasping when the index finger’s surface was visible (participants grasped the frustae with the index finger at the top). Our findings provide the first empirical report of the impact of asymmetric grasp surfaces on prehension and may have implications for scientists who wish to model reach-to-grasp behaviours.