Games are widely used as research environments for multi-agent reinforcement learning (MARL), but they pose three significant challenges: limited customization, high computational demands, and oversimplification. To address these issues, we introduce the first publicly available map editor for the popular mobile game Honor of Kings and design a lightweight environment, Mini Honor of Kings (Mini HoK), for researchers to conduct experiments. Mini HoK is highly efficient, allowing experiments to be run on personal PCs or laptops while still presenting sufficient challenges for existing MARL algorithms. We have tested our environment on common MARL algorithms and demonstrated that these algorithms have yet to find optimal solutions within this environment. This facilitates the dissemination and advancement of MARL methods within the research community. Additionally, we hope that more researchers will leverage the Honor of Kings map editor to develop innovative and scientifically valuable new maps. Our code and user manual are available at: https://github.com/tencent-ailab/mini-hok.

Recent pandemics have highlighted vulnerabilities in our global economic systems, especially supply chains. Possible future pandemic raises a dilemma for businesses owners between short-term profitability and long-term supply chain resilience planning. In this study, we propose a novel agent-based simulation model integrating extended Susceptible-Infected-Recovered (SIR) epidemiological model and supply and demand economic model to evaluate supply chain resilience strategies during pandemics. Using this model, we explore a range of supply chain resilience strategies under pandemic scenarios using in silico experiments. We find that a balanced approach to supply chain resilience performs better in both pandemic and non-pandemic times compared to extreme strategies, highlighting the importance of preparedness in the form of a better supply chain resilience. However, our analysis shows that the exact supply chain resilience strategy is hard to obtain for each firm and is relatively sensitive to the exact profile of the pandemic and economic state at the beginning of the pandemic. As such, we used a machine learning model that uses the agent-based simulation to estimate a near-optimal supply chain resilience strategy for a firm. The proposed model offers insights for policymakers and businesses to enhance supply chain resilience in the face of future pandemics, contributing to understanding the trade-offs between short-term gains and long-term sustainability in supply chain management before and during pandemics.

Training corpuses for vision language models (VLMs) typically lack sufficient amounts of decision-centric data. This renders off-the-shelf VLMs sub-optimal for decision-making tasks such as in-the-wild device control through graphical user interfaces (GUIs). While training with static demonstrations has shown some promise, we show that such methods fall short for controlling real GUIs due to their failure to deal with real-world stochasticity and non-stationarity not captured in static observational data. This paper introduces a novel autonomous RL approach, called DigiRL, for training in-the-wild device control agents through fine-tuning a pre-trained VLM in two stages: offline RL to initialize the model, followed by offline-to-online RL. To do this, we build a scalable and parallelizable Android learning environment equipped with a VLM-based evaluator and develop a simple yet effective RL approach for learning in this domain. Our approach runs advantage-weighted RL with advantage estimators enhanced to account for stochasticity along with an automatic curriculum for deriving maximal learning signal. We demonstrate the effectiveness of DigiRL using the Android-in-the-Wild (AitW) dataset, where our 1.3B VLM trained with RL achieves a 49.5% absolute improvement -- from 17.7 to 67.2% success rate -- over supervised fine-tuning with static human demonstration data. These results significantly surpass not only the prior best agents, including AppAgent with GPT-4V (8.3% success rate) and the 17B CogAgent trained with AitW data (38.5%), but also the prior best autonomous RL approach based on filtered behavior cloning (57.8%), thereby establishing a new state-of-the-art for digital agents for in-the-wild device control.

In this paper, we introduce novel gradient-based optimization methods for state-based potential games (SbPGs) within self-learning distributed production systems. SbPGs are recognised for their efficacy in enabling self-optimizing distributed multi-agent systems and offer a proven convergence guarantee, which facilitates collaborative player efforts towards global objectives. Our study strives to replace conventional ad-hoc random exploration-based learning in SbPGs with contemporary gradient-based approaches, which aim for faster convergence and smoother exploration dynamics, thereby shortening training duration while upholding the efficacy of SbPGs. Moreover, we propose three distinct variants for estimating the objective function of gradient-based learning, each developed to suit the unique characteristics of the systems under consideration. To validate our methodology, we apply it to a laboratory testbed, namely Bulk Good Laboratory Plant, which represents a smart and flexible distributed multi-agent production system. The incorporation of gradient-based learning in SbPGs reduces training times and achieves more optimal policies than its baseline.

Neural network supported tree-search has shown strong results in a variety of perfect information multi-agent tasks. However, the performance of these methods on partial information games has generally been below competing approaches. Here we study the class of simultaneous-move games, which are a subclass of partial information games which are most similar to perfect information games: both agents know the game state with the exception of the opponent's move, which is revealed only after each agent makes its own move. Simultaneous move games include popular benchmarks such as Google Research Football and Starcraft. In this study we answer the question: can we take tree search algorithms trained through self-play from perfect information settings and adapt them to simultaneous move games without significant loss of performance? We answer this question by deriving a practical method that attempts to approximate a coarse correlated equilibrium as a subroutine within a tree search. Our algorithm works on cooperative, competitive, and mixed tasks. Our results are better than the current best MARL algorithms on a wide range of accepted baseline environments.

Rapport is known as a conversational aspect focusing on relationship building, which influences outcomes in collaborative tasks. This study aims to establish human-agent rapport through small talk by using a rapport-building strategy. We implemented this strategy for the virtual agents based on dialogue strategies by prompting a large language model (LLM). In particular, we utilized two dialogue strategies-predefined sequence and free-form-to guide the dialogue generation framework. We conducted analyses based on human evaluations, examining correlations between total turn, utterance characters, rapport score, and user experience variables: naturalness, satisfaction, interest, engagement, and usability. We investigated correlations between rapport score and naturalness, satisfaction, engagement, and conversation flow. Our experimental results also indicated that using free-form to prompt the rapport-building strategy performed the best in subjective scores.

In decentralized multiagent trajectory planners, agents need to communicate and exchange their positions to generate collision-free trajectories. However, due to localization errors/uncertainties, trajectory deconfliction can fail even if trajectories are perfectly shared between agents. To address this issue, we first present PARM and PARM*, perception-aware, decentralized, asynchronous multiagent trajectory planners that enable a team of agents to navigate uncertain environments while deconflicting trajectories and avoiding obstacles using perception information. PARM* differs from PARM as it is less conservative, using more computation to find closer-to-optimal solutions. While these methods achieve state-of-the-art performance, they suffer from high computational costs as they need to solve large optimization problems onboard, making it difficult for agents to replan at high rates. To overcome this challenge, we present our second key contribution, PRIMER, a learning-based planner trained with imitation learning (IL) using PARM* as the expert demonstrator. PRIMER leverages the low computational requirements at deployment of neural networks and achieves a computation speed up to 5500 times faster than optimization-based approaches.

Evaluating autonomous vehicle stacks (AVs) in simulation typically involves replaying driving logs from real-world recorded traffic. However, agents replayed from offline data are not reactive and hard to intuitively control. Existing approaches address these challenges by proposing methods that rely on heuristics or generative models of real-world data but these approaches either lack realism or necessitate costly iterative sampling procedures to control the generated behaviours. In this work, we take an alternative approach and propose CtRL-Sim, a method that leverages return-conditioned offline reinforcement learning to efficiently generate reactive and controllable traffic agents. Specifically, we process real-world driving data through a physics-enhanced Nocturne simulator to generate a diverse offline reinforcement learning dataset, annotated with various reward terms. With this dataset, we train a return-conditioned multi-agent behaviour model that allows for fine-grained manipulation of agent behaviours by modifying the desired returns for the various reward components. This capability enables the generation of a wide range of driving behaviours beyond the scope of the initial dataset, including adversarial behaviours. We demonstrate that CtRL-Sim can generate diverse and realistic safety-critical scenarios while providing fine-grained control over agent behaviours.

Developing robots capable of executing various manipulation tasks, guided by natural language instructions and visual observations of intricate real-world environments, remains a significant challenge in robotics. Such robot agents need to understand linguistic commands and distinguish between the requirements of different tasks. In this work, we present Sigma-Agent, an end-to-end imitation learning agent for multi-task robotic manipulation. Sigma-Agent incorporates contrastive Imitation Learning (contrastive IL) modules to strengthen vision-language and current-future representations. An effective and efficient multi-view querying Transformer (MVQ-Former) for aggregating representative semantic information is introduced. Sigma-Agent shows substantial improvement over state-of-the-art methods under diverse settings in 18 RLBench tasks, surpassing RVT by an average of 5.2% and 5.9% in 10 and 100 demonstration training, respectively. Sigma-Agent also achieves 62% success rate with a single policy in 5 real-world manipulation tasks. The code will be released upon acceptance.

Recent studies have revealed that using social robots can accelerate the learning process of several skills in areas where autistic children typically show deficits. However, most early research studies conducted interactions via free play. More recent research has demonstrated that robot-mediated autism therapies focusing on core impairments of autism spectrum disorder (e.g., joint attention) yield better results than unstructured interactions. This paper aims to systematically review the most relevant findings concerning the application of social robotics to joint attention tasks, a cardinal feature of autism spectrum disorder that significantly influences the neurodevelopmental trajectory of autistic children. Initially, we define autism spectrum disorder and explore its societal implications. Following this, we examine the need for technological aid and the potentialities of robot-assisted autism therapy. We then define joint attention and highlight its crucial role in children's social and cognitive development. Subsequently, we analyze the importance of structured interactions and the role of selecting the optimal robot for specific tasks. This is followed by a comparative analysis of the works reviewed earlier, presenting an in-depth examination of two distinct formal models employed to design the prompts and reward system that enables the robot to adapt to children's responses. These models are critically compared to highlight their strengths and limitations. Next, we introduce a novel algorithm to address the identified limitations, integrating interactive environmental factors and a more sophisticated prompting and reward system. Finally, we propose further research directions, discuss the most relevant open questions, and draw conclusions regarding the effectiveness of social robotics in the medical treatment of autism spectrum disorders.