This work proposes a novel framework combining policy optimization with meta-learning to train highly adaptable MPC controllers, which enables rapid adaptation to unseen tasks, maintaining high performance at a fraction of the computational cost required for full retraining.
Abstract
Applying policy optimization to Model Predictive Control (MPC) yields high-performance and reliable controllers. However, the resulting controllers often overfit their training conditions and suffer significant performance degradation in unseen tasks. We propose a novel framework combining policy optimization with meta-learning to train highly adaptable MPC controllers. Our approach enables rapid adaptation to unseen tasks, maintaining high performance at a fraction of the computational cost required for full retraining. Furthermore, we integrate system identification into the pipeline to continuously refine both the MPC hyperparameters and the underlying predictive models. We validate our proposed methodology on a Ball-on-Plate system, demonstrating superior adaptability across various parameterized trajectory-tracking tasks.
This work presents a GPU-parallelized residual architecture that tightly integrates MPC and RL by blending their outputs at the torque-control level, achieving higher sample efficiency, converges to greater asymptotic rewards, expands the range of trackable velocity commands, and enables zero-shot adaptation to unseen...
Seungmin Jeon, Ho Jae Lee, Seung-Woo Hong et al.· IEEE Transactions on robotic...· 10 citations
A novel framework is proposed that integrates MPC with RL in a sequence decision-making framework and leverages a curvature-aware optimization to efficiently tackle non-convex loss landscapes and achieves higher returns and faster convergence.
Solver-Gradient Guided Reinforcement Learning is proposed, a solver-sensitivity augmentation for RL-based online MPC cost-weight adaptation that reaches PPO's best closed-loop return with up to 70.6% fewer samples, and outperforms GB-PL baselines by at least 54% in closed-loop return.
Baha Zarrouki, Arslan Thobani, Jasper Hoffmann et al.· 0 citations
A learning-based MPC framework that combines the data efficiency and structure of local model-based planning with learned components that compensate for incomplete dynamics and finite-horizon myopia and is integrated into a robotics simulator, enabling scalable model-based reinforcement learning under incomplete dynami...
. Soft robots, relying on compliant materials, exhibit inherent safety and environmental adaptability in complex environments, but their strong nonlinear characteristics make modeling and control particularly challenging. This paper employs a literature review and comparative analysis approach to summarize research on...
Hui-Yuan Liu· Proceedings of the 3rd Inter...· 0 citations
This paper studies learning-based synthesis of control barrier functions (CBFs) with an explicit treatment of hard input constraints. Training a task-specific CBF from scratch for each new environment can be computationally expensive and data-inefficient. To address this, we propose a meta-learning approach based on Mo...
Wataru Hashimoto, Masato Sakamoto, Kazumune Hashimoto et al.· Conference on Control Techno...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
