Reproduce “Towards Lifelong Federated Learning in Autonomous Mobile Robots with Continuous Sim-to-Real Transfer” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xianjia Yu, Jorge Pena Queralta, Tomi Westerlund
A framework for lifelong federated learning enables autonomous mobile robots to perform continuous vision-based obstacle avoidance by integrating data from both simulated and real robots. This approach demonstrated competitive accuracy against centralized training, improved sim-to-real knowledge transfer, and enhanced performance through collaborative model fusion.
Key insights
- Federated Learning offers robust and competitive performance for vision-based obstacle avoidance compared to centralized training, improving reliability.
- FL enhances the stability and effectiveness of sim-to-real knowledge transfer for robotic perception models.
- Automated data labeling, coupled with FL, enables a practical approach to continuous lifelong learning for robotic systems in dynamic environments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.