Event
Online Learning of Closed-Loop Dynamics Improves Mobile Manipulation Accuracy
1 reports · 1 sources
Tianlin Zhang and other authors propose a method for online learning of closed-loop dynamics under a frozen motion policy: a contaminated Gaussian model is used to distinguish predictable from unpredictable responses, and a Gaussian process combined with random Fourier features is used to build an online learner to reduce bias caused by heterogeneous observations. When combined with model predictive control (MPC), the method can predict floating-base motion more accurately. The authors report that in simulation and hardware experiments, across different manipulator configurations and tasks, the end-effector tracking error was 0.01 m, and they provide a finite-sample error bound for the online learning.
Report timeline
- Online Learning of Heterogeneous Locomotion Dynamics for Precise Mobile Manipulation
The study proposes an online method for learning closed-loop dynamics under a frozen locomotion policy, using a contaminated Gaussian model to distinguish predictable from unpredictable responses and combining a Gaussian process with random Fourier features to build an online learner that reduces bias from heterogeneous observations. Combined with model predictive control (MPC), the method predicts floating-base motion more accurately; in simulation and hardware experiments, end-effector tracking error was 0.01 m, and a finite-sample error bound for the online learning is provided.
arXiv cs.RO ·
Related topics
Interest trend
- Current
- 3
- Peak
- 4Nov 4 01:00
- Past 24 hours
- –
The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.