Continuously Improving Mobile Manipulation with Autonomous Real-World RL

Russell Mendonca, Emmanuel Panov, Bernadette Bucher, Jiuguang Wang, Deepak Pathak
Proceedings of The 8th Conference on Robot Learning, PMLR 270:5204-5219, 2025.

Abstract

We present a fully autonomous real-world RL framework for mobile manipulation that can learn policies without extensive instrumentation or human supervision. This is enabled by 1) task-relevant autonomy, which guides exploration towards object interactions and prevents stagnation near goal states, 2) efficient policy learning by leveraging basic task knowledge in behavior priors, and 3) formulating generic rewards that combine human-interpretable semantic information with low-level, fine-grained observations. We demonstrate that our approach allows Spot robots to continually improve their performance on a set of four challenging mobile manipulation tasks, obtaining an average success rate of 80% across tasks, a 3-4 times improvement over existing approaches. Videos can be found at https://continual-mobile-manip.github.io/.

Cite this Paper

BibTeX
@InProceedings{pmlr-v270-mendonca25a, title = {Continuously Improving Mobile Manipulation with Autonomous Real-World RL}, author = {Mendonca, Russell and Panov, Emmanuel and Bucher, Bernadette and Wang, Jiuguang and Pathak, Deepak}, booktitle = {Proceedings of The 8th Conference on Robot Learning}, pages = {5204--5219}, year = {2025}, editor = {Agrawal, Pulkit and Kroemer, Oliver and Burgard, Wolfram}, volume = {270}, series = {Proceedings of Machine Learning Research}, month = {06--09 Nov}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v270/main/assets/mendonca25a/mendonca25a.pdf}, url = {https://proceedings.mlr.press/v270/mendonca25a.html}, abstract = {We present a fully autonomous real-world RL framework for mobile manipulation that can learn policies without extensive instrumentation or human supervision. This is enabled by 1) task-relevant autonomy, which guides exploration towards object interactions and prevents stagnation near goal states, 2) efficient policy learning by leveraging basic task knowledge in behavior priors, and 3) formulating generic rewards that combine human-interpretable semantic information with low-level, fine-grained observations. We demonstrate that our approach allows Spot robots to continually improve their performance on a set of four challenging mobile manipulation tasks, obtaining an average success rate of 80% across tasks, a 3-4 times improvement over existing approaches. Videos can be found at https://continual-mobile-manip.github.io/.} }
Endnote
%0 Conference Paper %T Continuously Improving Mobile Manipulation with Autonomous Real-World RL %A Russell Mendonca %A Emmanuel Panov %A Bernadette Bucher %A Jiuguang Wang %A Deepak Pathak %B Proceedings of The 8th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Pulkit Agrawal %E Oliver Kroemer %E Wolfram Burgard %F pmlr-v270-mendonca25a %I PMLR %P 5204--5219 %U https://proceedings.mlr.press/v270/mendonca25a.html %V 270 %X We present a fully autonomous real-world RL framework for mobile manipulation that can learn policies without extensive instrumentation or human supervision. This is enabled by 1) task-relevant autonomy, which guides exploration towards object interactions and prevents stagnation near goal states, 2) efficient policy learning by leveraging basic task knowledge in behavior priors, and 3) formulating generic rewards that combine human-interpretable semantic information with low-level, fine-grained observations. We demonstrate that our approach allows Spot robots to continually improve their performance on a set of four challenging mobile manipulation tasks, obtaining an average success rate of 80% across tasks, a 3-4 times improvement over existing approaches. Videos can be found at https://continual-mobile-manip.github.io/.
APA
Mendonca, R., Panov, E., Bucher, B., Wang, J. & Pathak, D.. (2025). Continuously Improving Mobile Manipulation with Autonomous Real-World RL. Proceedings of The 8th Conference on Robot Learning, in Proceedings of Machine Learning Research 270:5204-5219 Available from https://proceedings.mlr.press/v270/mendonca25a.html.

Related Material