# Integrated Conflict Management for UAM with Strategic Demand Capacity Balancing and Learning-based Tactical Deconfliction

**Authors:** Shulu Chen, Antony Evans, Marc Brittain, Peng Wei  
**arXiv:** [2305.10556](https://arxiv.org/abs/2305.10556)  
**Date:** 2023-05-17  
**Summary page:** https://www.summarizepaper.com/en/paper/2305.10556v1/

## Key points

- Urban air mobility (UAM) has the potential to revolutionize daily transportation in urban environments
- Aviation safety is crucial for commercializing UAM
- Strategic and tactical deconfliction is necessary to separate aircraft safely
- Reinforcement learning has shown promise in tactically deconflicting en route commercial air traffic
- Performance of reinforcement learning depends on traffic density
- Authors propose a framework that combines demand capacity balancing (DCB) with reinforcement learning
- DCB preconditioning enables reinforcement learning to achieve better performance in tactical safety separation
- Results show that DCB allows target safety levels to be met which would otherwise be impossible
- Combining strategic DCB with reinforcement learning achieves greater operational efficiency compared to alternatives
- Integration of strategic conflict management and tactical deconfliction is important for safe implementation of UAM systems.

## Summary

Urban air mobility (UAM) has the potential to revolutionize daily transportation by providing efficient passenger and cargo deliveries within urban environments. Ensuring aviation safety is crucial before commercializing this emerging transportation mode. This requires safely separating all aircraft through strategic and tactical deconfliction. Reinforcement learning has shown promise in tactically deconflicting en route commercial air traffic in simulations; however, its performance is dependent on traffic density. To address this issue, the authors propose a novel framework that combines demand capacity balancing (DCB) for strategic conflict management with reinforcement learning for tactical separation. The framework leverages DCB to precondition traffic to appropriate density levels, enabling reinforcement learning to achieve significantly better performance in tactical safety separation. The results demonstrate that DCB preconditioning allows target safety levels to be met which would otherwise be impossible. Moreover, combining strategic DCB with reinforcement learning not only meets these safety levels but also achieves greater operational efficiency compared to alternative solutions. This research highlights the importance of integrating strategic conflict management and tactical deconfliction techniques for UAM's safe implementation. By combining DCB and reinforcement learning, this framework offers a promising approach to ensure aviation safety while optimizing operational efficiency in urban air mobility systems.