Learning geometry-aware virtual guides from few demonstration points
Learning Geometry-Aware Virtual Fixtures From Sparse Demonstrations
Robotics
Summary
Teleoperated robots help people perform tasks remotely, but teaching them precise paths can require many examples, which is hard to collect. The authors propose a method that uses just a few key points to teach these paths, while also guiding the robot's orientation using math that respects 3D geometry. Their approach adjusts how strongly the robot follows the guide depending on confidence, allowing the operator to work freely outside defined regions. They demonstrate their method on simple examples and a precise surgical cutting task using a robotic surgery platform.
What this means in practice
- •For robotic surgery teams: Create precise motion guides from a few demonstration points to assist surgical tool manipulation with adapted force feedback.$Commercial implications: Enables surgical robot systems to offer customizable, geometry-aware virtual fixtures that improve operator ease and task precision.
- •For industrial robot operators: Set up guidance paths quickly for teleoperated robots performing precise assembly or cutting tasks with limited demonstration data.
Authors
Maximilian Mühlbauer, Marcella Piacentino, Raffaella Mancino, Paolino De Risi, Bernhard Weber, Margarida Campos, Thomas Hulin, Sylvain Calinon, Freek Stulp, Alin Albu-Schäffer, Julian Klodmann, Fanny Ficuciello, João Silvério
Abstract
In many teleoperation applications, collecting a large number of demonstrations as required for traditional probabilistic learning from demonstration (LfD) approaches may not be feasible. To still give operators the ability to intuitively create trajectories as Virtual Fixtures (VFs), we propose to leverage a motion prior in the learning process. Particularly, by using Linear Quadratic Tracking (LQT), users are able to define guiding trajectories from the demonstration of just a few via points. To account for orientation guidance, we further reformulate classical LQT on Riemannian manifolds, introducing an additional geometric prior. Through a probabilistic interpretation of the LQT solution, we derive a covariance estimate at each trajectory point which we use to modulate the stiffness of the resulting fixture, resulting in strong guidance around the via points and softer guidance when far away. The covariance information is also used to define a validity region of the fixture, allowing the operator to leave its influence area and conduct unmodeled tasks. We evaluate the proposed Riemannian LQT formulation in a set of toy examples and the full framework on a cutting task requiring high precision in a minimally invasive surgery setting on the da Vinci Research Kit (dVRK).