Differentiable Iterative Learning Control for Black-Box Position-Commanded Manipulators
Overview
This research introduces Differentiable Iterative Learning Control (Diff-ILC), an online bilevel framework that adapts constrained iterative learning control parameters between physical robot executions without requiring access to low-level controllers. By identifying a closed-loop differential model from paired smooth perturbations and using hypergradients derived from implicitly differentiating the locally active KKT system, Diff-ILC enables multiple computational parameter updates without additional robot executions. Experiments on a 6-DoF DOBOT CR3 show that after only two trial-to-trial updates, Diff-ILC achieves 78.8% and 76.4% lower mean joint-vector and Cartesian RMSE, respectively, compared to fixed-parameter model-based ILC across four primary trajectories.
Generated from attributed reports · Updated 1 hours ago
Event evidence and corrections
0 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Data origin: robot trajectories
“enabling multiple computational parameter updates without additional robot executions. Experiments on a 6-DoF DOBOT CR3 across unseen planar and three-dimensional trajectories show substantially faster physical-trial convergence than fixed-parameter model-based ILC. Across the four primary trajectories, after only two trial-to-trial updates, Diff-ILC achieves 78.8% and 76.4% lower mean joint-vector and Cartesian RMSE”
Exact source · revision 1Source owner not reported
Reported quantity · trajectories: 4 other · reported trials
“enabling multiple computational parameter updates without additional robot executions. Experiments on a 6-DoF DOBOT CR3 across unseen planar and three-dimensional trajectories show substantially faster physical-trial convergence than fixed-parameter model-based ILC. Across the four primary trajectories, after only two trial-to-trial updates, Diff-ILC achieves 78.8% and 76.4% lower mean joint-vector and Cartesian RMSE”
Exact source · revision 1Source owner not reported
Reported quantity · trial-to-trial updates: 2 other · Basis not reported
“enabling multiple computational parameter updates without additional robot executions. Experiments on a 6-DoF DOBOT CR3 across unseen planar and three-dimensional trajectories show substantially faster physical-trial convergence than fixed-parameter model-based ILC. Across the four primary trajectories, after only two trial-to-trial updates, Diff-ILC achieves 78.8% and 76.4% lower mean joint-vector and Cartesian RMSE”
Exact source · revision 1Source owner not reported
Reported quantity · DoF: 6 other · Basis not reported
“enabling multiple computational parameter updates without additional robot executions. Experiments on a 6-DoF DOBOT CR3 across unseen planar and three-dimensional trajectories show substantially faster physical-trial convergence than fixed-parameter model-based ILC. Across the four primary trajectories, after only two trial-to-trial updates, Diff-ILC achieves 78.8% and 76.4% lower mean joint-vector and Cartesian RMSE”
Exact source · revision 1Source owner not reported
Report timeline
Follow attributed reports and material updates.
- arXiv Robotics — research abstractsSignalDifferentiable Iterative Learning Control for Black-Box Position-Commanded Manipulators
This research introduces Differentiable Iterative Learning Control (Diff-ILC), an online bilevel framework that adapts constrained iterative learning control parameters between physical robot executions without requiring access to low-level controllers. By identifying a closed-loop differential model from paired smooth perturbations and using hypergradients derived from implicitly differentiating the locally active KKT system, Diff-ILC enables multiple computational parameter updates without additional robot executions. Experiments on a 6-DoF DOBOT CR3 show that after only two trial-to-trial updates, Diff-ILC achieves 78.8% and 76.4% lower mean joint-vector and Cartesian RMSE, respectively, compared to fixed-parameter model-based ILC across four primary trajectories.
Event coverage history
There is not enough continuous observation data to show a trend.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).