A collaborative distributed multi-agent reinforcement learning technique for dynamic agent shortest path planning via selected sub-goals in complex cluttered environments

Megherbi, Dalila B.; Kim, M.

doi:10.1109/cogsima.2015.7108185

Cited by 5 publications

(3 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The scheme adopts and extends prior studies [1,7,24] to demonstrate the agent learning process in a distributed environment with one or more sub-goals, where multiple agents have different final goals (destinations) for agent path planning. In prior studies, RL was mainly used as the agent learning process to selfimprove learning performance [7,10,14,23]. RL is also the study of machine learning algorithms to automatically attempt and find maximizing cumulative rewards for faster optimal path planning in terms of value and policy networks.…”

Section: A Contributions and Objectivesmentioning

confidence: 99%

See 1 more Smart Citation

Autonomous and Asynchronous Triggered Agent Exploratory Path-planning Via a Terrain Clutter-index using Reinforcement Learning

Kim

2022

JICCE

View full text Add to dashboard Cite

An intelligent distributed multi-agent system (IDMS) using reinforcement learning (RL) is a challenging and intricate problem in which single or multiple agent(s) aim to achieve their specific goals (sub-goal and final goal), where they move their states in a complex and cluttered environment. The environment provided by the IDMS provides a cumulative optimal reward for each action based on the policy of the learning process. Most actions involve interacting with a given IDMS environment; therefore, it can provide the following elements: a starting agent state, multiple obstacles, agent goals, and a cluttered index. The reward in the environment is also reflected by RL-based agents, in which agents can move randomly or intelligently to reach their respective goals, to improve the agent learning performance. We extend different cases of intelligent multi-agent systems from our previous works: (a) a proposed environment-clutter-based-index for agent sub-goal selection and analysis of its effect, and (b) a newly proposed RL reward scheme based on the environmental clutter-index to identify and analyze the prerequisites and conditions for improving the overall system.

show abstract

Section: A Contributions and Objectivesmentioning

confidence: 99%

“…The proposed method has been used to develop time-varying [12,20] and real-time applications [21] such as mobile robotics. The RL-based multi-agent approach can counter many different problems, such as machine learning, to solve multi-agent coordination and collaboration [22,23].…”

Section: Introductionmentioning

confidence: 99%

Autonomous and Asynchronous Triggered Agent Exploratory Path-planning Via a Terrain Clutter-index using Reinforcement Learning

Kim

2022

JICCE

View full text Add to dashboard Cite

show abstract

“…Constraints imposed due to the terrain make the translation of such algorithms for ground vehicles inefficient. Algorithms have been proposed for collaborative path finding for autonomous vehicles [17][18][19][20]. For many critical missions like transportation of military cargo, personnel and mined ore, it is desirable that the vehicles are commandeered real time by humans.…”

Section: Introductionmentioning

confidence: 99%

A Hierarchical Route Guidance Framework for Off-Road Connected Vehicles

Roy

Wan²,

Goswami

et al. 2018

Journal of Dynamic Systems, Measurement, and Control

View full text Add to dashboard Cite

A new framework for route guidance, as part of a path decision support tool, for off-road driving scenarios is presented in this paper. The algorithm accesses information gathered prior to and during a mission which are stored as layers of a central map. The algorithm incorporates a priori knowledge of the low resolution soil and elevation information and real-time high-resolution information from on-board sensors. The challenge of high computational cost to find the optimal path over a large-scale high-resolution map is mitigated by the proposed hierarchical path planning algorithm. A dynamic programming (DP) method generates the globally optimal path approximation based on low-resolution information. The optimal cost-to-go from each grid cell to the destination is calculated by back-stepping from the target and stored. A model predictive control algorithm (MPC) operates locally on the vehicle to find the optimal path over a moving radial horizon. The MPC algorithm uses the stored global optimal cost-to-go map in addition to high resolution and locally available information. Efficacy of the developed algorithm is demonstrated in scenarios simulating static and moving obstacles avoidance, path finding in condition-time-variant environments, eluding adversarial line of sight detection, and connected fleet cooperation.

show abstract

Multi-step Reinforcement Learning Algorithm of Mobile Robot Path Planning Based on Virtual Potential Field

Liu

Wei

2017

Communications in Computer and Information Science

View full text Add to dashboard Cite

A collaborative distributed multi-agent reinforcement learning technique for dynamic agent shortest path planning via selected sub-goals in complex cluttered environments

Cited by 5 publications

References 29 publications

Autonomous and Asynchronous Triggered Agent Exploratory Path-planning Via a Terrain Clutter-index using Reinforcement Learning

Autonomous and Asynchronous Triggered Agent Exploratory Path-planning Via a Terrain Clutter-index using Reinforcement Learning

A Hierarchical Route Guidance Framework for Off-Road Connected Vehicles

Multi-step Reinforcement Learning Algorithm of Mobile Robot Path Planning Based on Virtual Potential Field

Contact Info

Product

Resources

About