Multi-Agent Reinforcement Learning (MARL) has emerged as a foundational approach for addressing diverse, intelligent control tasks, notably in autonomous driving within the Internet of Vehicles (IoV) domain. However, the widely assumed existence of a central node for centralized, federated learning-assisted MARL might be impractical in highly dynamic environments. This can lead to excessive communication overhead, potentially overwhelming the IoV system. To address these challenges, we design a novel communication-efficient and policy collaboration algorithm for MARL under the frameworks of Soft Actor-Critic (SAC) and Decentralized Federated Learning (DFL), named RSM-MASAC, within a fully distributed architecture. In particular, RSM-MASAC enhances multi-agent collaboration and prioritizes higher communication efficiency in dynamic IoV system by incorporating the concept of segmented aggregation in DFL and augmenting multiple model replicas from received neighboring policy segments, which are subsequently employed as reconstructed referential policies for mixing. Distinctively diverging from traditional RL approaches, with derived new bounds under Maximum Entropy Reinforcement Learning (MERL), RSM-MASAC adopts a theory-guided mixture metric to regulate the selection of contributive referential policies to guarantee the soft policy improvement during communication phase. Finally, the extensive simulations in mixed-autonomy traffic control scenarios verify the effectiveness and superiority of our algorithm.

Peer learning is a novel high-level reinforcement learning framework for agents learning in groups. While standard reinforcement learning trains an individual agent in trial-and-error fashion, all on its own, peer learning addresses a related setting in which a group of agents, i.e., peers, learns to master a task simultaneously together from scratch. Peers are allowed to communicate only about their own states and actions recommended by others: "What would you do in my situation?". Our motivation is to study the learning behavior of these agents. We formalize the teacher selection process in the action advice setting as a multi-armed bandit problem and therefore highlight the need for exploration. Eventually, we analyze the learning behavior of the peers and observe their ability to rank the agents' performance within the study group and understand which agents give reliable advice. Further, we compare peer learning with single agent learning and a state-of-the-art action advice baseline. We show that peer learning is able to outperform single-agent learning and the baseline in several challenging discrete and continuous OpenAI Gym domains. Doing so, we also show that within such a framework complex policies from action recommendations beyond discrete action spaces can evolve.

We present a modular approach to \emph{reinforcement learning} (RL) in environments consisting of simpler components evolving in parallel. A monolithic view of such modular environments may be prohibitively large to learn, or may require unrealizable communication between the components in the form of a centralized controller. Our proposed approach is based on the assume-guarantee paradigm where the optimal control for the individual components is synthesized in isolation by making \emph{assumptions} about the behaviors of neighboring components, and providing \emph{guarantees} about their own behavior. We express these \emph{assume-guarantee contracts} as regular languages and provide automatic translations to scalar rewards to be used in RL. By combining local probabilities of satisfaction for each component, we provide a lower bound on the probability of satisfaction of the complete system. By solving a Markov game for each component, RL can produce a controller for each component that maximizes this lower bound. The controller utilizes the information it receives through communication, observations, and any knowledge of a coarse model of other agents. We experimentally demonstrate the efficiency of the proposed approach on a variety of case studies.

The rise of AI-based and autonomous systems is raising concerns and apprehension due to potential negative repercussions stemming from their behavior or decisions. These systems must be designed to comply with the human contexts in which they will operate. To this extent, Townsend et al. (2022) introduce the concept of SLEEC (social, legal, ethical, empathetic, or cultural) rules that aim to facilitate the formulation, verification, and enforcement of the rules AI-based and autonomous systems should obey. They lay out a methodology to elicit them and to let philosophers, lawyers, domain experts, and others to formulate them in natural language. To enable their effective use in AI systems, it is necessary to translate these rules systematically into a formal language that supports automated reasoning. In this study, we first conduct a linguistic analysis of the SLEEC rules pattern, which justifies the translation of SLEEC rules into classical logic. Then we investigate the computational complexity of reasoning about SLEEC rules and show how logical programming frameworks can be employed to implement SLEEC rules in practical scenarios. The result is a readily applicable strategy for implementing AI systems that conform to norms expressed as SLEEC rules.

In the Multiagent Path Finding problem (MAPF for short), we focus on efficiently finding non-colliding paths for a set of $k$ agents on a given graph $G$, where each agent seeks a path from its source vertex to a target. An important measure of the quality of the solution is the length of the proposed schedule $\ell$, that is, the length of a longest path (including the waiting time). In this work, we propose a systematic study under the parameterized complexity framework. The hardness results we provide align with many heuristics used for this problem, whose running time could potentially be improved based on our fixed-parameter tractability results. We show that MAPF is W[1]-hard with respect to $k$ (even if $k$ is combined with the maximum degree of the input graph). The problem remains NP-hard in planar graphs even if the maximum degree and the makespan$\ell$ are fixed constants. On the positive side, we show an FPT algorithm for $k+\ell$. As we delve further, the structure of~$G$ comes into play. We give an FPT algorithm for parameter $k$ plus the diameter of the graph~$G$. The MAPF problem is W[1]-hard for cliquewidth of $G$ plus $\ell$ while it is FPT for treewidth of $G$ plus $\ell$.

Learning to collaborate has witnessed significant progress in multi-agent reinforcement learning (MARL). However, promoting coordination among agents and enhancing exploration capabilities remain challenges. In multi-agent environments, interactions between agents are limited in specific situations. Effective collaboration between agents thus requires a nuanced understanding of when and how agents' actions influence others. To this end, in this paper, we propose a novel MARL algorithm named Situation-Dependent Causal Influence-Based Cooperative Multi-agent Reinforcement Learning (SCIC), which incorporates a novel Intrinsic reward mechanism based on a new cooperation criterion measured by situation-dependent causal influence among agents. Our approach aims to detect inter-agent causal influences in specific situations based on the criterion using causal intervention and conditional mutual information. This effectively assists agents in exploring states that can positively impact other agents, thus promoting cooperation between agents. The resulting update links coordinated exploration and intrinsic reward distribution, which enhance overall collaboration and performance. Experimental results on various MARL benchmarks demonstrate the superiority of our method compared to state-of-the-art approaches.

Building general-purpose robots that can operate seamlessly, in any environment, with any object, and utilizing various skills to complete diverse tasks has been a long-standing goal in Artificial Intelligence. Unfortunately, however, most existing robotic systems have been constrained - having been designed for specific tasks, trained on specific datasets, and deployed within specific environments. These systems usually require extensively-labeled data, rely on task-specific models, have numerous generalization issues when deployed in real-world scenarios, and struggle to remain robust to distribution shifts. Motivated by the impressive open-set performance and content generation capabilities of web-scale, large-capacity pre-trained models (i.e., foundation models) in research fields such as Natural Language Processing (NLP) and Computer Vision (CV), we devote this survey to exploring (i) how these existing foundation models from NLP and CV can be applied to the field of robotics, and also exploring (ii) what a robotics-specific foundation model would look like. We begin by providing an overview of what constitutes a conventional robotic system and the fundamental barriers to making it universally applicable. Next, we establish a taxonomy to discuss current work exploring ways to leverage existing foundation models for robotics and develop ones catered to robotics. Finally, we discuss key challenges and promising future directions in using foundation models for enabling general-purpose robotic systems. We encourage readers to view our living GitHub repository of resources, including papers reviewed in this survey as well as related projects and repositories for developing foundation models for robotics.

This paper introduces a new approach to solving a continuous-time version of the multi-agent path finding problem. The algorithm translates the problem into an extension of the classical Boolean satisfiability problem, satisfiability modulo theories (SMT), that can be solved by off-the-shelf solvers. This enables the exploitation of conflict generalization techniques that such solvers can handle. Computational experiments show that the new approach scales better with respect to the available computation time than state-of-the art approaches and is usually able to avoid their exponential behavior on a class of benchmark problems modeling a typical bottleneck situation.

Among the great successes of Reinforcement Learning (RL), self-play algorithms play an essential role in solving competitive games. Current self-play algorithms optimize the agent to maximize expected win-rates against its current or historical copies, making it often stuck in the local optimum and its strategy style simple and homogeneous. A possible solution is to improve the diversity of policies, which helps the agent break the stalemate and enhances its robustness when facing different opponents. However, enhancing diversity in the self-play algorithms is not trivial. In this paper, we aim to introduce diversity from the perspective that agents could have diverse risk preferences in the face of uncertainty. Specifically, we design a novel reinforcement learning algorithm called Risk-sensitive Proximal Policy Optimization (RPPO), which smoothly interpolates between worst-case and best-case policy learning and allows for policy learning with desired risk preferences. Seamlessly integrating RPPO with population-based self-play, agents in the population optimize dynamic risk-sensitive objectives with experiences from playing against diverse opponents. Empirical results show that our method achieves comparable or superior performance in competitive games and that diverse modes of behaviors emerge. Our code is public online at \url{https://github.com/Jackory/RPBT}.

Decision makers often aim to learn a treatment assignment policy under a capacity constraint on the number of agents that they can treat. When agents can respond strategically to such policies, competition arises, complicating estimation of the optimal policy. In this paper, we study capacity-constrained treatment assignment in the presence of such interference. We consider a dynamic model where the decision maker allocates treatments at each time step and heterogeneous agents myopically best respond to the previous treatment assignment policy. When the number of agents is large but finite, we show that the threshold for receiving treatment under a given policy converges to the policy's mean-field equilibrium threshold. Based on this result, we develop a consistent estimator for the policy gradient. In simulations and a semi-synthetic experiment with data from the National Education Longitudinal Study of 1988, we demonstrate that this estimator can be used for learning capacity-constrained policies in the presence of strategic behavior.