Generating Reactive Virtual Guidance Fixtures for Assisted Telemanipulation Tasks

In this work, we propose the generation of a tube-shaped virtual fixture that guides the operator towards a target position. This position is obtained through a HumanMachine Interface (HMI) and it is used to adapt the endpoint of the virtual fixture by means of a reactive constraint-based approach. In parallel, a target object is inferred in realtime through a Bayesian intent estimation algorithm that uses information coming from a vision system and the HMI. A shared control strategy is then used to attract the endpoint of the virtual fixture towards the inferred target object. The path of this reactive virtual fixture also adapts to avoid collisions with dynamic obstacles. With our approach, we expect to reduce the effort and fatigue of the operators when performing telemanipulation tasks.