Dual Control Reference Generation for Single-Shot Robotic Task Execution under Parametric Uncertainty

Victor Vantilborgh, Tom Lefebvre, Guillaume Crevecoeur, Hrishikesh Sathyanarayan, Ian Abraham

Keywords: Dual Control, Uncertainty-Aware Control, Active System Identification, Trajectory Generation

Abstract: We consider the problem of single-shot robotic task execution under parametric uncertainty: a partially unknown robot must complete a task accurately on its first attempt, without a prior identification phase. We propose a generic framework for reference trajectory generation that deliberately excites only task-relevant parameters during execution, enabling concurrent online identification and control. We propose to solve a robust trajectory optimization problem, integrating adaptive closed-loop dynamics with an ad-hoc adaptation law, and use the reference as an artificial input. We show that our formulation naturally reasons about Fisher information. We validated the solution on a 7-DoF Franka FR3 pick-and-place task. Our results suggest that control-aware trajectories yield significantly improved task accuracy and identification compared to conventional approaches.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 8