1/ 🚗 🌏 What if an autonomous vehicle could move to a new city without collecting a single human demonstration in that city? I am so excited to introduce our new work: Learning to Drive in New Cities Without Human Demonstrations.
Image from Zilin Wang (@zilinwang4ai)
1
11
8
48
18,598
2/ 🤔 Autonomous vehicles now outperform humans within specific operating regions, but their deployment to new cities remains costly and slow. Key bottleneck: Collecting human demonstrations from new cities.
1
4
482
3/ 💡 However, the lane-level map and traffic meta-information are prevalent and inexpensive. Our solution: Adapting driving policies to new cities by map-based self-play multi-agent reinforcement learning. ❌ Target-city human trajectories. ❌ Expensive data collection loop.
Image from Zilin Wang (@zilinwang4ai)
1
4
393
4/ 🚄 We introduce NO data Map-based self-play for Autonomous Driving (NOMAD), which adapts a driving policy to a new city using only the city's lane-level map + meta info. The policy is trained via KL-regularized self-play in a target-city simulator.
Image from Zilin Wang (@zilinwang4ai)

Feb 20, 2026 · 3:00 PM UTC

2
4
328
RelevantRecentLikes
5/ 🚀 The result (closed-loop Boston → Singapore): - Success: 42% → >90% - Realism (WOSAC): 0.679 → ~0.765 The resulting policies form an upper envelope that Pareto-dominates baselines. We use a simple reward function and the same hyperparameters for all city pairs.
Image from Zilin Wang (@zilinwang4ai)
1
4
289
6/ 🔬 We also analyze: • role of behavioral priors • necessity of target-city maps • comparison to methods that do use target-city demos • generalization across cities • sensitivity to KL strength • evaluation under non-self-play agents • effect of map mirroring
Image from Zilin Wang (@zilinwang4ai)
1
5
295
7/ 🔮 NOMAD substantially narrows cross-city generalization gaps, supporting scalable deployment of autonomous driving systems across diverse environments and highlighting the promise of self-play MARL for improving safety and robustness. Paper: arxiv.org/abs/2602.15891
Learning to Drive in New Cities Without Human Demonstrations

Learning to Drive in New Cities Without Human Demonstrations

While autonomous vehicles have achieved reliable performance within specific operating regions, their deployment to new cities remains costly and slow. A key bottleneck is the need to collect many...

arxiv.org
1
1
3
556
8/ ❤️ Thanks to my amazing collaborators @saeedrmd @daphne_cor @bidiptas13 @AlexDGoldie @j_foerst @shimon8282. Thanks to all colleagues for the helpful discussions. If you’re into AVs / RL / ML, we’d love your thoughts!
1
6
516