Autonomous Separation Assurance with Deep Multi-Agent Reinforcement Learning

Brittain, Marc; Yang, Xuxi; Wei, Peng

doi:10.2514/1.i010973

Cited by 20 publications

(11 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed framework uses a customized version of proximal policy optimization incorporating an attention network to achieve high traffic throughput under uncertainty, allowing agents to access different aircraft information in the sector. During the training process, a single neural network is taught and shared by all agents, employing a centralized learning and decentralized execution model [ 12 ]. To validate the proposed framework, the authors utilize three case studies in the BlueSky air traffic simulator.…”

Section: Literature Surveymentioning

confidence: 99%

A study on multi-agent reinforcement learning for autonomous distribution vehicles

Ergün

2023

Iran J Comput Sci

View full text Add to dashboard Cite

A self-driving automobile, also known as an autonomous vehicle, can observe its surroundings and maneuver without the assistance of a driver thanks to software algorithms and a completely automated driving system. As a result, the car may react to the environment in a way that is comparable to a human driver. Road regulations and limitations are required to guarantee the security and effectiveness of delivery services, as well as to stop accidents and damage brought on by technology failures. This paper formulates the Autonomous Delivery Vehicles optimization problem and proposes a Multi-Agent Reinforcement Learning approach that makes use of shortest-path data that have been obtained analytically. This method allows for the training of multiple agents to work together and improve delivery. Using this strategy, autonomous delivery vehicles' efficiency and security can be markedly enhanced.

show abstract

Section: Literature Surveymentioning

confidence: 99%

A study on multi-agent reinforcement learning for autonomous distribution vehicles

Ergün

2023

Iran J Comput Sci

View full text Add to dashboard Cite

show abstract

“…The learning-based aviation system under test is a DRL model presented in [21,35], which was designed for aircraft separation assurance. In this proposed DRL framework, individual aircraft are represented by agents in en-route airspace, allowing autonomous separation from other aircraft.…”

Section: Runtime Bootstrappingmentioning

confidence: 99%

A Verification Framework for Certifying Learning-Based Safety-Critical Aviation Systems

Baheri¹,

Ren²,

Johnson³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

We present a safety verification framework for design-time and run-time assurance of learning-based components in aviation systems. Our proposed framework integrates two novel methodologies. From the design-time assurance perspective, we propose offline mixedfidelity verification tools that incorporate knowledge from different levels of granularity in simulated environments. From the run-time assurance perspective, we propose reachabilityand statistics-based online monitoring and safety guards for a learning-based decision-making model to complement the offline verification methods. This framework is designed to be loosely coupled among modules, allowing the individual modules to be developed using independent methodologies and techniques, under varying circumstances and with different tool access. The proposed framework offers feasible solutions for meeting system safety requirements at different stages throughout the system development and deployment cycle, enabling the continuous learning and assessment of the system product.

show abstract

“…For the purpose of identifying and resolving conflicts among a variable number of airplanes in a high-density, stochastic, and dynamic in route sector, a deep multi-agent RL framework is suggested in Brittain et al (2021). A customized version of proximal policy optimization that incorporates an attention network is used in the proposed framework.…”

Section: Literature Surveymentioning

confidence: 99%

A Study on Multi-agent Reinforcement Learning for Autonomous Distribution Vehicles

Ergün

2022

Preprint

View full text Add to dashboard Cite

A self-driving car, also called an autonomous vehicle, is capable of sensing the environment and driving without any human intervention, utilizing software algorithms and a fully automated driving system. This allows the vehicle to manage external conditions similar to a human. Safe and efficient delivery services require road controls and road restrictions to prevent accidents and reduce damage from unexpected technical failures. This study formulates the Autonomous Delivery Vehicles optimization problem and proposes a multi-agent reinforcement learning method using the analytically calculated shortest path information.

show abstract

Autonomous Separation Assurance with Deep Multi-Agent Reinforcement Learning

Cited by 20 publications

References 23 publications

A study on multi-agent reinforcement learning for autonomous distribution vehicles

A study on multi-agent reinforcement learning for autonomous distribution vehicles

A Verification Framework for Certifying Learning-Based Safety-Critical Aviation Systems

A Study on Multi-agent Reinforcement Learning for Autonomous Distribution Vehicles

Contact Info

Product

Resources

About