Unified Meta-Representation and Feedback Calibration for General Disturbance Estimation
Zihan Yang, Jindou Jia, Meng Wang, Yuhang Liu, Kexin Guo, Xiang Yu
AI summary
Problem
Existing meta-learning-based disturbance estimators rely on shared structural representations that fail for unstructured, time-varying disturbances. Furthermore, representation errors and distribution shifts during online adaptation severely degrade prediction accuracy.
Approach
The framework extracts a unified representation from a finite time window of past state observations without predefined structural assumptions, then applies a feedback-calibrated online adaptation mechanism to correct prediction errors and attenuate learning residuals.
Key results
- Unified meta-representation learned from finite-time historical states without structural assumptions
- Feedback-calibrated online adaptation that attenuates learning residuals and improves generalization
- Theoretical proof of simultaneous exponential convergence for learning and estimation errors
- Superior trajectory tracking and disturbance rejection on a quadrotor under rapidly changing, non-structural disturbances
Why it matters
Enables reliable, adaptive disturbance rejection for complex robotic systems operating in unpredictable, unstructured environments where traditional model-based or meta-learning methods fail.
Abstract
Precise control in modern robotic applications is always an open issue due to unknown time-varying dis- turbances. Existing meta-learning-based approaches require a shared representation of environmental structures, which lack flexibility for realistic non-structural disturbances. Besides, representation error and the distribution shifts can lead to heavy degradation in prediction accuracy. This work presents a generalizable disturbance estimation framework that builds on meta-learning and feedback-calibrated online adaptation. By extracting features from a finite time window of past observations, a unified representation that effectively captures general non-structural disturbances can be learned without predefined structural assumptions. The online adaptation pro- cess is subsequently calibrated by a state-feedback mechanism to attenuate the learning residual originating from the repre- sentation and generalizability limitations. Theoretical analysis shows that simultaneous convergence of both the online learning error and the disturbance estimation error can be achieved. Through the unified meta-representation, our framework ef- fectively estimates multiple rapidly changing disturbances, as demonstrated by quadrotor flight experiments. See the project page for video, supplementary material and code: https: //nonstructural-metalearn.github.io.