A-NC: Adaptive Neural Control with implicit online inference of privileged parameters

Marcin Paluch, Florian Bolli, Pehuen Moure, Xiang Deng, Tobi Delbruck
Proceedings of the 7th Annual Learning for Dynamics \& Control Conference, PMLR 283:987-998, 2025.

Abstract

Rapid changes in the environment and robot parameters pose significant challenges for control systems, particularly when key parameters are not directly measurable. In this paper, we introduce a novel approach using classical Recurrent Neural Network (RNN) controllers to dynamically adapt policies in response to these changes. We propose strategies for data collection and processing that enable the successful training of efficient Gated Recurrent Unit (GRU) nonlinear controllers capable of adapting to changing parameters. We demonstrate this approach using a simulated and a physical cartpole robot. The RNNs are trained through supervised learning on data generated in simulation using Nonlinear Model Predictive Control (NMPC). We vary the cartpole’s angle sensor offset or pole length jointly with pole mass, none of which are directly measurable by the robot. Our results show how the RNN controller adjusts its policy based on past trajectories, leading to control that mimics the NMPC, outperforming Domain Randomization (DR) technique applied to feedforward neural networks. Unlike NMPC, which relies on explicit knowledge of environment parameters, the RNN implicitly estimates these parameters from past trajectories, allowing it to adapt its control policy dynamically. It also outperforms NMPC control performance when the parameters relevant for NMPC are not known.

Cite this Paper

BibTeX
@InProceedings{pmlr-v283-paluch25a, title = {A-NC: Adaptive Neural Control with implicit online inference of privileged parameters}, author = {Paluch, Marcin and Bolli, Florian and Moure, Pehuen and Deng, Xiang and Delbruck, Tobi}, booktitle = {Proceedings of the 7th Annual Learning for Dynamics \& Control Conference}, pages = {987--998}, year = {2025}, editor = {Ozay, Necmiye and Balzano, Laura and Panagou, Dimitra and Abate, Alessandro}, volume = {283}, series = {Proceedings of Machine Learning Research}, month = {04--06 Jun}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v283/main/assets/paluch25a/paluch25a.pdf}, url = {https://proceedings.mlr.press/v283/paluch25a.html}, abstract = {Rapid changes in the environment and robot parameters pose significant challenges for control systems, particularly when key parameters are not directly measurable. In this paper, we introduce a novel approach using classical Recurrent Neural Network (RNN) controllers to dynamically adapt policies in response to these changes. We propose strategies for data collection and processing that enable the successful training of efficient Gated Recurrent Unit (GRU) nonlinear controllers capable of adapting to changing parameters. We demonstrate this approach using a simulated and a physical cartpole robot. The RNNs are trained through supervised learning on data generated in simulation using Nonlinear Model Predictive Control (NMPC). We vary the cartpole’s angle sensor offset or pole length jointly with pole mass, none of which are directly measurable by the robot. Our results show how the RNN controller adjusts its policy based on past trajectories, leading to control that mimics the NMPC, outperforming Domain Randomization (DR) technique applied to feedforward neural networks. Unlike NMPC, which relies on explicit knowledge of environment parameters, the RNN implicitly estimates these parameters from past trajectories, allowing it to adapt its control policy dynamically. It also outperforms NMPC control performance when the parameters relevant for NMPC are not known.} }
Endnote
%0 Conference Paper %T A-NC: Adaptive Neural Control with implicit online inference of privileged parameters %A Marcin Paluch %A Florian Bolli %A Pehuen Moure %A Xiang Deng %A Tobi Delbruck %B Proceedings of the 7th Annual Learning for Dynamics \& Control Conference %C Proceedings of Machine Learning Research %D 2025 %E Necmiye Ozay %E Laura Balzano %E Dimitra Panagou %E Alessandro Abate %F pmlr-v283-paluch25a %I PMLR %P 987--998 %U https://proceedings.mlr.press/v283/paluch25a.html %V 283 %X Rapid changes in the environment and robot parameters pose significant challenges for control systems, particularly when key parameters are not directly measurable. In this paper, we introduce a novel approach using classical Recurrent Neural Network (RNN) controllers to dynamically adapt policies in response to these changes. We propose strategies for data collection and processing that enable the successful training of efficient Gated Recurrent Unit (GRU) nonlinear controllers capable of adapting to changing parameters. We demonstrate this approach using a simulated and a physical cartpole robot. The RNNs are trained through supervised learning on data generated in simulation using Nonlinear Model Predictive Control (NMPC). We vary the cartpole’s angle sensor offset or pole length jointly with pole mass, none of which are directly measurable by the robot. Our results show how the RNN controller adjusts its policy based on past trajectories, leading to control that mimics the NMPC, outperforming Domain Randomization (DR) technique applied to feedforward neural networks. Unlike NMPC, which relies on explicit knowledge of environment parameters, the RNN implicitly estimates these parameters from past trajectories, allowing it to adapt its control policy dynamically. It also outperforms NMPC control performance when the parameters relevant for NMPC are not known.
APA
Paluch, M., Bolli, F., Moure, P., Deng, X. & Delbruck, T.. (2025). A-NC: Adaptive Neural Control with implicit online inference of privileged parameters. Proceedings of the 7th Annual Learning for Dynamics \& Control Conference, in Proceedings of Machine Learning Research 283:987-998 Available from https://proceedings.mlr.press/v283/paluch25a.html.

Related Material