Decentralized Data Collection for Robotic Fleet Learning: A Game-Theoretic Approach

Oguzhan Akcin, Po-han Li, Shubhankar Agarwal, Sandeep P. Chinchali
Proceedings of The 6th Conference on Robot Learning, PMLR 205:978-988, 2023.

Abstract

Fleets of networked autonomous vehicles (AVs) collect terabytes of sensory data, which is often transmitted to central servers (the “cloud”) for training machine learning (ML) models. Ideally, these fleets should upload all their data, especially from rare operating contexts, in order to train robust ML models. However, this is infeasible due to prohibitive network bandwidth and data labeling costs. Instead, we propose a cooperative data sampling strategy where geo-distributed AVs collaborate to collect a diverse ML training dataset in the cloud. Since the AVs have a shared objective but minimal information about each other’s local data distribution and perception model, we can naturally cast cooperative data collection as an $N$-player mathematical game. We show that our cooperative sampling strategy uses minimal information to converge to a centralized oracle policy with complete information about all AVs. Moreover, we theoretically characterize the performance benefits of our game-theoretic strategy compared to greedy sampling. Finally, we experimentally demonstrate that our method outperforms standard benchmarks by up to $21.9%$ on 4 perception datasets, including for autonomous driving in adverse weather conditions. Crucially, our experimental results on real-world datasets closely align with our theoretical guarantees.

Cite this Paper


BibTeX
@InProceedings{pmlr-v205-akcin23a, title = {Decentralized Data Collection for Robotic Fleet Learning: A Game-Theoretic Approach}, author = {Akcin, Oguzhan and Li, Po-han and Agarwal, Shubhankar and Chinchali, Sandeep P.}, booktitle = {Proceedings of The 6th Conference on Robot Learning}, pages = {978--988}, year = {2023}, editor = {Liu, Karen and Kulic, Dana and Ichnowski, Jeff}, volume = {205}, series = {Proceedings of Machine Learning Research}, month = {14--18 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v205/akcin23a/akcin23a.pdf}, url = {https://proceedings.mlr.press/v205/akcin23a.html}, abstract = {Fleets of networked autonomous vehicles (AVs) collect terabytes of sensory data, which is often transmitted to central servers (the “cloud”) for training machine learning (ML) models. Ideally, these fleets should upload all their data, especially from rare operating contexts, in order to train robust ML models. However, this is infeasible due to prohibitive network bandwidth and data labeling costs. Instead, we propose a cooperative data sampling strategy where geo-distributed AVs collaborate to collect a diverse ML training dataset in the cloud. Since the AVs have a shared objective but minimal information about each other’s local data distribution and perception model, we can naturally cast cooperative data collection as an $N$-player mathematical game. We show that our cooperative sampling strategy uses minimal information to converge to a centralized oracle policy with complete information about all AVs. Moreover, we theoretically characterize the performance benefits of our game-theoretic strategy compared to greedy sampling. Finally, we experimentally demonstrate that our method outperforms standard benchmarks by up to $21.9%$ on 4 perception datasets, including for autonomous driving in adverse weather conditions. Crucially, our experimental results on real-world datasets closely align with our theoretical guarantees.} }
Endnote
%0 Conference Paper %T Decentralized Data Collection for Robotic Fleet Learning: A Game-Theoretic Approach %A Oguzhan Akcin %A Po-han Li %A Shubhankar Agarwal %A Sandeep P. Chinchali %B Proceedings of The 6th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2023 %E Karen Liu %E Dana Kulic %E Jeff Ichnowski %F pmlr-v205-akcin23a %I PMLR %P 978--988 %U https://proceedings.mlr.press/v205/akcin23a.html %V 205 %X Fleets of networked autonomous vehicles (AVs) collect terabytes of sensory data, which is often transmitted to central servers (the “cloud”) for training machine learning (ML) models. Ideally, these fleets should upload all their data, especially from rare operating contexts, in order to train robust ML models. However, this is infeasible due to prohibitive network bandwidth and data labeling costs. Instead, we propose a cooperative data sampling strategy where geo-distributed AVs collaborate to collect a diverse ML training dataset in the cloud. Since the AVs have a shared objective but minimal information about each other’s local data distribution and perception model, we can naturally cast cooperative data collection as an $N$-player mathematical game. We show that our cooperative sampling strategy uses minimal information to converge to a centralized oracle policy with complete information about all AVs. Moreover, we theoretically characterize the performance benefits of our game-theoretic strategy compared to greedy sampling. Finally, we experimentally demonstrate that our method outperforms standard benchmarks by up to $21.9%$ on 4 perception datasets, including for autonomous driving in adverse weather conditions. Crucially, our experimental results on real-world datasets closely align with our theoretical guarantees.
APA
Akcin, O., Li, P., Agarwal, S. & Chinchali, S.P.. (2023). Decentralized Data Collection for Robotic Fleet Learning: A Game-Theoretic Approach. Proceedings of The 6th Conference on Robot Learning, in Proceedings of Machine Learning Research 205:978-988 Available from https://proceedings.mlr.press/v205/akcin23a.html.

Related Material