Abstract:

Humans excel at composing high-level plans that achieve a complex, multimodal objective; however, achieving proficiency in teleoperating multi-rotor aerial vehicles (MAVs) in unstructured environments with stability and safety requires significant skill and training. In this talk, we present human-in-the-loop control of a MAV via teleoperation using motion primitives that addresses these concerns. We show that we can increase naive user proficiency by utilizing known vehicle models. We remove the requirement of maintaining stability and dynamic feasibility from the operator by generating snap-continuous motion primitives at user specified transition points. We further provide safety by incorporating reactive collision avoidance via input space search with locally generated depth maps using onboard depth cameras. Motion primitives based teleoperation is readily extendable to shared control, e.g. through sampling-based adaptation based on local directional intent prediction over motion primitive libraries.

Speaker Bio:

Xuning Yang is a Ph.D. student in the Resilient Intelligent Systems Lab at the Robotics Institute, advised by Prof. Nathan Michael. Her research focuses on enabling efficient human-in-the-loop control for mobile robots, including online modelling and prediction of operator intent in order to enable agile safe navigation in unstructured environments. Prior to CMU, She received her B.A.Sc. in Engineering Science from University of Toronto, majoring in Aerospace Engineering.