Decentralized Non-communicating Multiagent Collision Avoidance with Deep Reinforcement Learning.

This work presents a decentralized multiagent collision avoidance algorithm based on a novel application of deep reinforcement learning, which effectively offloads the online computation (for predicting ... Finding feasible, collision-free paths for multiagent systems can be challenging, particularly in non-communicating scenarios where each agent's intent (e.g. goal) is unobservable to the others. ... Sequential Decision Making A non-communicating multiagent collision avoidance problem can be formulated as a partially-observable sequential decision making problem. ...

arXiv:1609.07845v2 fatcat:e4njfuo56ng5vmgfy5576jurdy

Multiple Versions

This work presents a decentralized multiagent collision avoidance algorithm based on a novel application of deep reinforcement learning, which effectively offloads the online computation (for predicting ... Simulation results show more than 26% improvement in paths quality (i.e., time to reach the goal) when compared with optimal reciprocal collision avoidance (ORCA), a state-of-the-art collision avoidance ... Sequential Decision Making A non-communicating multiagent collision avoidance problem can be formulated as a partially-observable sequential decision making problem. ...

doi:10.1109/icra.2017.7989037 dblp:conf/icra/ChenLEH17 fatcat:duqucw3ab5btdm5idvpaljrqqm

We survey alternative approaches to conflict resolution that rely on emerging technologies such as deep learning. ... Robot Path Planning and Collision Avoidance Agents with shared resources within a multiagent system, can benefit from prioritization [2] . ... Characteristics of a Multiagent environment include [2, 18] ; • Multiagent environments, specify interaction protocols and communication protocols. • Environment is usually open and decentralized. • Agents ...

doi:10.1088/1742-6596/1235/1/012022 fatcat:nvu3mbdxzra5rnyeju4di2gi7i

Open Access

In [16] , a multiagent deep reinforcement learning (MDRL) approach is proposed with a focus on the scalability and learning stability: IQL is implemented with an inter-agent shared fingerprint in the ... Thus, Deep Reinforcement Learning (DRL) has become the most common way to deal with the function approximation (neural networks as non-linear parametric approximators). ...

doi:10.1109/access.2021.3053348 fatcat:hht7ikou65cqtk6q62j4leq3oi

DOAJ

We establish a safe MARL framework with decentralized multiple CBFs and develop Multi-Agent Deep Deterministic Policy Gradient (MADDPG) to Multi-Agent Deep Deterministic Policy Gradient with decentralized ... Based on a collision-avoidance problem that includes not only cooperative agents but obstacles, we demonstrate the construction of multiple CBFs with safety guarantees in theory. ... Multi-Agent Reinforcement Learning MARL scales RL to environments with multiple agents. ...

arXiv:2103.12553v1 fatcat:dqowkp4kojcntemtm5hsyy46rq

Zhang, H., +, TCYB Oct. 2020 4268-4280 Collision avoidance A Robust Collision Perception Visual Neural Network With Specific Selec- tivity to Darker Objects. ... ., +, TCYB Oct. 2020 4481-4494 Decision making Deep Reinforcement Learning for Multiagent Systems: A Review of Challenges, Solutions, and Applications. ...

doi:10.1109/tcyb.2020.3047216 fatcat:5giw32c2u5h23fu4drupnh644a

We present a novel algorithm (DeepMNavigate) for global multi-agent navigation in dense scenarios using deep reinforcement learning (DRL). ... We highlight the algorithm's benefits over prior learning methods and geometric decentralized algorithms in complex scenarios. ... [16] , [17] describe a decentralized multi-robot collision avoidance framework where each robot makes navigation decisions independently without any communication with other agents. ...

arXiv:1910.09441v5 fatcat:yf5y45lcxze33h54isuvpvspom

Multiple Versions

We use a decentralized form of the actor-critic approach to deep reinforcement learning$-$multi-agent deep deterministic policy gradient. ... In our framework, we employed an actor-critic architecture with a centralized critic and decentralized actors to avoid the problem of a non-stationary environment [36] induced by decentralized learning ... However, in large problems with many state-action pairs, to avoid Bellman's "curse of dimensionality," deep reinforcement learning methods (DRL), such as Deep Q-network (DQN) [28] , are used where the ...

arXiv:2109.11672v2 fatcat:sod5cbu3fjg7bfl2dc7moc7tqy

Multiple Versions

We focus on decentralized navigation among multiple non-communicating rational agents at uncontrolled intersections, i.e., street intersections without traffic signs or signals. ... Avoiding collisions in such domains relies on the ability of agents to predict each others' intentions reliably, and react quickly. ... Some works employ implicit models of interaction, such as risk level sets [25] , deep reinforcement learning [4] , or imitation learning [26] . ...

arXiv:2011.03894v1 fatcat:62dp2zntfvhuhhji6yv44bfvoa

We focus on navigation among rational, non-communicating agents at unsignalized street intersections. ... We design a decentralized planning algorithm that generates actions aimed at reducing the uncertainty over the mode of the emerging multiagent behavior. ... [15] learn a policy for crossing unsignalized intersections under occlusions using deep reinforcement learning and show how it outperforms selected rule-based baselines. Finally, Okamoto et al. ...

arXiv:2004.05205v2 fatcat:hgkomx2eqfajdkwzkwzweyixv4

Multiple Versions

To reduce online computation, we first decompose the multi-agent scenario and solve a two agents collision avoidance problem using reinforcement learning (RL). ... In this paper, we propose a fast multi-agent collision avoidance algorithm for general nonlinear agents with continuous action space, where each agent observes only positions and velocities of nearby agents ... Two Agents Collision Avoidance via Reinforcement Learning Network Structure: For general policy-based RL algorithm, the policy is usually represented as a neural network, with the state as input and the ...

arXiv:1910.10887v2 fatcat:5ygtfqrjrrfnfnpn33vumumsqe

Multiple Versions

Multi-Agent Reinforcement Learning has arisen as a powerful method to accomplish this task because it considers the interaction between agents and also allows for decentralized training -- which makes ... We show that the rewards of the transferred policies with MAPPO and domain randomization are, on average, 1.85 times superior to the rule-based method. ... Multi-Agent Deep Reinforcement Learning The MARL problem can be modeled as a Decentralized Partially Observable MDP (Dec-POMDP), which can be defined by the tuple N, S, A i , T, R, Ω i , O , where N is ...

arXiv:2203.11653v1 fatcat:b6pb4cu72ncgdha4gl7gzshrdq

Open Access

., +, TNNLS Collision avoidance Nopoles, G., +, TNNLS March 2020 865-875 Nonsynaptic Error Backpropagation in Long-Term Cognitive Networks. ... ., +, TNNLS Oct. 2020 4381-4388 AlphaSeq: Sequence Discovery With Deep Reinforcement Learning. ...

doi:10.1109/tnnls.2020.3045307 fatcat:34qoykdtarewhdscxqj5jvovqy

INDEX TERMS Collision avoidance, deep reinforcement learning, motion planning, multiagent systems, decentralized execution. ... This work proposes using deep reinforcement (RL) learning as a framework to model the complex interactions and cooperation with nearby, decision-making agents, such as pedestrians and other robots. ... ACKNOWLEDGMENT Yu Fan Chen was with the Massachusetts Institute of Technology (MIT), Cambridge, MA 02139, USA. ...

doi:10.1109/access.2021.3050338 fatcat:k54zg6mip5ganahre4jpianh3y

DOAJ

We validate the learned deep neural network policy in a set of simulated and real scenarios with noisy measurements and demonstrate that our method is able to generate a robust navigation strategy that ... We train the network on a large number of frames of collision avoidance data collected by repeatedly running a multi-agent simulator with different parameter settings. ... Machine Learning for Multi-Agent Systems Reinforcement learning has been widely used for the multiagent decision making [13] - [16] , which is formulated as a multi-agent Markov Decision Processes (MDP ...

arXiv:1609.06838v2 fatcat:2n74xhmenfg5vop5lifa4bf6yy

Multiple Versions

Decentralized Non-communicating Multiagent Collision Avoidance with Deep Reinforcement Learning [article]

Preserved Fulltext

Decentralized non-communicating multiagent collision avoidance with deep reinforcement learning

Preserved Fulltext

Conflict resolution via emerging technologies?

Preserved Fulltext

A Multiagent Deep Reinforcement Learning Approach for Path Planning in Autonomous Surface Vehicles: The YpacaraC-Lake Patrolling Case

Preserved Fulltext

Safe Multi-Agent Reinforcement Learning through Decentralized Multiple Control Barrier Functions [article]

Preserved Fulltext

2020 Index IEEE Transactions on Cybernetics Vol. 50

Preserved Fulltext

DeepMNavigate: Deep Reinforced Multi-Robot Navigation Unifying Local Global Collision Avoidance [article]

Preserved Fulltext

Other Versions

A Multi-Agent Deep Reinforcement Learning Coordination Framework for Connected and Automated Vehicles at Merging Roadways [article]

Preserved Fulltext

Other Versions

Multimodal Trajectory Prediction via Topological Invariance for Navigation at Uncontrolled Intersections [article]

Preserved Fulltext

Implicit Multiagent Coordination at Unsignalized Intersections via Multimodal Inference Enabled by Topological Braids [article]

Preserved Fulltext

Other Versions

Reciprocal Collision Avoidance for General Nonlinear Agents using Reinforcement Learning [article]

Preserved Fulltext

Other Versions

Transferring Multi-Agent Reinforcement Learning Policies for Autonomous Driving using Sim-to-Real [article]

Preserved Fulltext

2020 Index IEEE Transactions on Neural Networks and Learning Systems Vol. 31

Preserved Fulltext

Collision Avoidance in Pedestrian-Rich Environments with Deep Reinforcement Learning

Preserved Fulltext

Deep-Learned Collision Avoidance Policy for Distributed Multi-Agent Navigation [article]

Preserved Fulltext

Other Versions