# Zero-Shot Autonomous Vehicle Policy Transfer: From Simulation to Real-World via Adversarial Learning

arXiv: https://arxiv.org/abs/1903.05252 · HTML: https://mlsanity.com/paper/1903.05252/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Mar 2019.

## Built on by (most cited first, 7 of 7)

- [Challenges and Opportunities in Deep Reinforcement Learning with Graph Neural Networks: A Comprehensive review of Algorithms and Applications](https://mlsanity.com/paper/2206.07922/) (2022)
- [Evaluating the Robustness of Collaborative Agents](https://mlsanity.com/paper/2101.05507/) (2021)
- [Adversarial Deep Reinforcement Learning for Improving the Robustness of Multi-agent Autonomous Driving Policies](https://mlsanity.com/paper/2112.11937/) (2021)
- [SemTra: A Semantic Skill Translator for Cross-Domain Zero-Shot Policy Adaptation](https://mlsanity.com/paper/2402.07418/) (2024)
- [Suicidal Pedestrian: Generation of Safety-Critical Scenarios for Autonomous Vehicles](https://mlsanity.com/paper/2309.00249/) (2023)
- [Exiting the Simulation: The Road to Robust and Resilient Autonomous Vehicles at Scale](https://mlsanity.com/paper/2210.10876/) (2022)
- [ReMAV: Reward Modeling of Autonomous Vehicles for Finding Likely Failure Events](https://mlsanity.com/paper/2308.14550/) (2023)

## Abstract (arXiv)

In this article, we demonstrate a zero-shot transfer of an autonomous driving policy from simulation to University of Delaware's scaled smart city with adversarial multi-agent reinforcement learning, in which an adversary attempts to decrease the net reward by perturbing both the inputs and outputs of the autonomous vehicles during training. We train the autonomous vehicles to coordinate with each other while crossing a roundabout in the presence of an adversary in simulation. The adversarial policy successfully reproduces the simulated behavior and incidentally outperforms, in terms of travel time, both a human-driving baseline and adversary-free trained policies. Finally, we demonstrate that the addition of adversarial training considerably improves the performance eatstability and robustness of the policies after transfer to the real world compared to Gaussian noise injection.