# HetGPS: Scalable Graph Multi-Agent Reinforcement Learning with Physics-Anchored Adaptive Safety for EV Charging

arXiv: https://arxiv.org/abs/2608.00679 · HTML: https://mlsanity.com/paper/2608.00679/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 39 references that are in the map.
- First posted on arXiv 1 Aug 2026.

## Builds on (most cited first, 11 of 11)

- [Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor](https://mlsanity.com/paper/1801.01290/) (2018)
- [Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments](https://mlsanity.com/paper/1706.02275/) (2017)
- [Recipe for a General, Powerful, Scalable Graph Transformer](https://mlsanity.com/paper/2205.12454/) (2022)
- [Safe Reinforcement Learning via Shielding](https://mlsanity.com/paper/1708.08611/) (2017)
- [Responsive Safety in Reinforcement Learning by PID Lagrangian Methods](https://mlsanity.com/paper/2007.03964/) (2020)
- [Graph Convolutional Reinforcement Learning](https://mlsanity.com/paper/1810.09202/) (2018)
- [A predictive safety filter for learning-based control of constrained nonlinear dynamical systems](https://mlsanity.com/paper/1812.05506/) (2018)
- [Optimal Power Flow Using Graph Neural Networks](https://mlsanity.com/paper/1910.09658/) (2019)
- [Reinforcement Learning with Adaptive Regularization for Safe Control of Critical Systems](https://mlsanity.com/paper/2404.15199/) (2024)
- [Safety Constrained Multi-Agent Reinforcement Learning for Active Voltage Control](https://mlsanity.com/paper/2405.08443/) (2024)
- [Towards Efficient Collaboration via Graph Modeling in Reinforcement Learning](https://mlsanity.com/paper/2410.15841/) (2024)

## Abstract (arXiv)

Safety interventions for large populations of network-coupled agents must protect shared constraints without unnecessarily overriding task-oriented policy decisions. We present HetGPS, a hybrid graph-control framework synergizing learned graph risk with physics-anchored correction by separating intervention magnitude from corrective direction. An action-conditioned graph residual model schedules state-dependent intervention authority, while a physics model determines its direction. For electric vehicle (EV) charging, we couple this filter with a parameter-shared heterogeneous graph soft actor-critic policy, enabling topology-aware coordination with a learned model size independent of fleet size. Across five nested distribution networks with 200--3,218 EVs and 100 evaluation days, Adaptive Authority reduces bus--step voltage violations from 3.93--7.74% without filtering to 0.52--3.44%, while maintaining 99.06--100% departure success. Relative to the same physics-directed projection with fixed authority, it improves mean reward on all five networks and lowers the mean safety score on four. The deployed policy-and-risk model contains 383,702 learned parameters at every scale; at 3,218 EVs, a matched centralized SAC actor is about 170× larger. A policy trained on the eight-transformer system transfers zero-shot to the 16- and 32-transformer systems, attaining 0.57--0.75% violation rates and at least 99.99% departure success. These results show that learned graph risk can allocate intervention authority at scale while feeder physics anchors corrective action.