aiqbench analysis
This research presents a bi-level optimization framework that simultaneously calibrates noise covariances and kinematic parameters for legged robots, addressing the common issue of manually tuned estimators. By differentiating through an estimator-in-the-loop system, the method achieves significantly improved state estimation accuracy on quadrupedal and humanoid platforms without relying on hand-tuned baselines. The work advances data-driven robotics by unifying sensor calibration with kinematic modeling into a single principled framework for dynamic environments.
Simultaneous Calibration of Noise Covariance and Kinematics for State Estimation of Legged Robots via Bi-level Optimization
arXiv:2510.11539v5 Announce Type: replace Abstract: Accurate state estimation is critical for legged and aerial robots operating in dynamic, uncertain environments. A key challenge lies in specifying process and measurement noise covariances, which are typically unknown or manually tuned. In this work, we introduce a bi-level optimization framework that jointly calibrates covariance matrices and kinematic parameters in an estimator-in-the-loop manner. The upper level treats noise covariances and model parameters as optimization variables, while the lower level executes a full-information estimator. Differentiating through the estimator allows direct optimization of trajectory-level objectives, resulting in accurate and consistent state estimates. We validate our approach on quadrupedal and humanoid robots, demonstrating significantly improved estimation accuracy and uncertainty calibration compared to hand-tuned baselines. Our method unifies state estimation, sensor, and kinematics calibration into a principled, data-driven framework applicable across diverse robotic platforms.
Abstract:Accurate state estimation is critical for legged and aerial robots operating in dynamic, uncertain environments. A key challenge lies in specifying process and measurement noise covariances, which are typically unknown or manually tuned. In this work, we introduce a bi-level optimization framework that jointly calibrates covariance matrices and kinematic parameters in an estimator-in-the-loop manner. The upper level treats noise covariances and model parameters as optimization variables, while the lower level executes a full-information estimator. Differentiating through the estimator allows direct optimization of trajectory-level objectives, resulting in accurate and consistent state estimates. We validate our approach on quadrupedal and humanoid robots, demonstrating significantly improved estimation accuracy and uncertainty calibration compared to hand-tuned baselines. Our method unifies state estimation, sensor, and kinematics calibration into a principled, data-driven framework applicable across diverse robotic platforms.
Robotics (cs.RO); Optimization and Control (math.OC)
arXiv:2510.11539 [cs.RO]
(or arXiv:2510.11539v5 [cs.RO] for this version)
Full text under the publisher's open licence · 145 words
Related coverage
Limitations on Joint Coherence Transfer in Quantum Thermodynamics
arXiv · Sep 16, 2026
Continuous Policy and Value Iteration for Relaxed Stochastic Control Problems and Its Convergence
arXiv · Oct 5, 2026
Exact Unitary Synthesis: Simultaneously Optimal in Size and Depth-Ancilla Tradeoff
arXiv · Oct 7, 2026
Simultaneously Query-Optimal Quantum Linear-System Algorithm
arXiv · Sep 30, 2026