Despite impressive successes, deep reinforcement learning (RL) systems still fall short of human performance on generalization to new tasks and environments that differ from their training. As a benchmark tailored for studying RL generalization, we introduce Avalon, a set of tasks in which embodied agents in highly diverse procedural 3D worlds must survive by navigating terrain, hunting or gathering food, and avoiding hazards. Avalon is unique among existing RL benchmarks in that the reward function, world dynamics, and action space are the same for every task, with tasks differentiated solely by altering the environment; its 20 tasks, ranging in complexity from eat and throw to hunt and navigate, each create worlds in which the agent must perform specific skills in order to survive. This setup enables investigations of generalization within tasks, between tasks, and to compositional tasks that require combining skills learned from previous tasks. Avalon includes a highly efficient simulator, a library of baselines, and a benchmark with scoring metrics evaluated against hundreds of hours of human performance, all of which are open-source and publicly available. We find that standard RL baselines make progress on most tasks but are still far from human performance, suggesting Avalon is challenging enough to advance the quest for generalizable RL.

This paper describes an agent-based model of epidemics dynamics. This model is willingly simplified, as its goal is not to predict the evolution of the epidemics, but to explain the underlying mechanisms in an interactive way. This model allows to compare screening prioritisation strategies, as well as vaccination priority strategies, on a virtual population. The model is implemented in Netlogo in different simulators, published online to let people experiment with them. This paper reports on the model design, implementation, and experimentations. In particular we have compared screening strategies to evaluate the epidemics vs control it by quarantining infectious people; and we have compared vaccinating older people with more risk factors, vs younger people with more social contacts.

This paper investigates when one can efficiently recover an approximate Nash Equilibrium (NE) in offline congestion games.The existing dataset coverage assumption in offline general-sum games inevitably incurs a dependency on the number of actions, which can be exponentially large in congestion games. We consider three different types of feedback with decreasing revealed information. Starting from the facility-level (a.k.a., semi-bandit) feedback, we propose a novel one-unit deviation coverage condition and give a pessimism-type algorithm that can recover an approximate NE. For the agent-level (a.k.a., bandit) feedback setting, interestingly, we show the one-unit deviation coverage condition is not sufficient. On the other hand, we convert the game to multi-agent linear bandits and show that with a generalized data coverage assumption in offline linear bandits, we can efficiently recover the approximate NE. Lastly, we consider a novel type of feedback, the game-level feedback where only the total reward from all agents is revealed. Again, we show the coverage assumption for the agent-level feedback setting is insufficient in the game-level feedback setting, and with a stronger version of the data coverage assumption for linear bandits, we can recover an approximate NE. Together, our results constitute the first study of offline congestion games and imply formal separations between different types of feedback.

The pursuit-evasion game in Smart City brings a profound impact on the Multi-vehicle Pursuit (MVP) problem, when police cars cooperatively pursue suspected vehicles. Existing studies on the MVP problems tend to set evading vehicles to move randomly or in a fixed prescribed route. The opponent modeling method has proven considerable promise in tackling the non-stationary caused by the adversary agent. However, most of them focus on two-player competitive games and easy scenarios without the interference of environments. This paper considers a Team-to-Team Multi-vehicle Pursuit (T2TMVP) problem in the complicated urban traffic scene where the evading vehicles adopt the pre-trained dynamic strategies to execute decisions intelligently. To solve this problem, we propose an opponent-aware reinforcement learning via maximizing mutual information indicator (OARLM2I2) method to improve pursuit efficiency in the complicated environment. First, a sequential encoding-based opponents joint strategy modeling (SEOJSM) mechanism is proposed to generate evading vehicles' joint strategy model, which assists the multi-agent decision-making process based on deep Q-network (DQN). Then, we design a mutual information-united loss, simultaneously considering the reward fed back from the environment and the effectiveness of opponents' joint strategy model, to update pursuing vehicles' decision-making process. Extensive experiments based on SUMO demonstrate our method outperforms other baselines by 21.48% on average in reducing pursuit time. The code is available at \url{https://github.com/ANT-ITS/OARLM2I2}.

Let $L$ be a finite lattice and $\mathcal{E}(L)$ be the set of join endomorphisms of $L$. We consider the problem of given $L$ and $f,g \in \mathcal{E}(L)$, finding the greatest lower bound $f \sqcap_{{\scriptsize \mathcal{E}(L)}} g$ in the lattice $\mathcal{E}(L)$. (1) We show that if $L$ is distributive, the problem can be solved in time $O(n)$ where $n=| L |$. The previous upper bound was $O(n^2)$. (2) We provide new algorithms for arbitrary lattices and give experimental evidence that they are significantly faster than the existing algorithm. (3) We characterize the standard notion of distributed knowledge of a group as the greatest lower bound of the join-endomorphisms representing the knowledge of each member of the group. (4) We show that deciding whether an agent has the distributed knowledge of two other agents can be computed in time $O(n^2)$ where $n$ is the size of the underlying set of states. (5) For the special case of $S5$ knowledge, we show that it can be decided in time $O(n\alpha_{n})$ where $\alpha_{n}$ is the inverse of the Ackermann function.

Adaptive curricula in reinforcement learning (RL) have proven effective for producing policies robust to discrepancies between the train and test environment. Recently, the Unsupervised Environment Design (UED) framework generalized RL curricula to generating sequences of entire environments, leading to new methods with robust minimax regret properties. Problematically, in partially-observable or stochastic settings, optimal policies may depend on the ground-truth distribution over aleatoric parameters of the environment in the intended deployment setting, while curriculum learning necessarily shifts the training distribution. We formalize this phenomenon as curriculum-induced covariate shift (CICS), and describe how its occurrence in aleatoric parameters can lead to suboptimal policies. Directly sampling these parameters from the ground-truth distribution avoids the issue, but thwarts curriculum learning. We propose SAMPLR, a minimax regret UED method that optimizes the ground-truth utility function, even when the underlying training data is biased due to CICS. We prove, and validate on challenging domains, that our approach preserves optimality under the ground-truth distribution, while promoting robustness across the full range of environment settings.

I show that when the money supply accelerates, the learning agents only adjust their actions, which include consumption and demand for real balance, after gathering learning experience for many periods. This delayed adjustments leads to low returns during transition periods. Once they start adjusting to the new environment, their welfare improves. Their changes in beliefs and actions lead to temporary inflation volatility. I also show that, 1. the AI agents who explores their environment more adapt to the policy regime change quicker, which leads to welfare improvements and less inflation volatility, and 2. the AI agents who have experienced a structural change adjust their beliefs and behaviours quicker than an inexperienced learning agent.

Evaluation of intervention in a multi-agent system, e.g., when humans should intervene in autonomous driving systems and when a player should pass to teammates for a good shot, is challenging in various engineering and scientific fields. Estimating the individual treatment effect (ITE) using counterfactual long-term prediction is practical to evaluate such interventions. However, most of the conventional frameworks did not consider the time-varying complex structure of multi-agent relationships and covariate counterfactual prediction. This may sometimes lead to erroneous assessments of ITE and interpretation problems. Here we propose an interpretable, counterfactual recurrent network in multi-agent systems to estimate the effect of the intervention. Our model leverages graph variational recurrent neural networks and theory-based computation with domain knowledge for the ITE estimation framework based on long-term prediction of multi-agent covariates and outcomes, which can confirm under the circumstances under which the intervention is effective. On simulated models of an automated vehicle and biological agents with time-varying confounders, we show that our methods achieved lower estimation errors in counterfactual covariates and the most effective treatment timing than the baselines. Furthermore, using real basketball data, our methods performed realistic counterfactual predictions and evaluated the counterfactual passes in shot scenarios.

Intelligence agents and multi-agent systems play important roles in scenes like the control system of grouped drones, and multi-agent navigation and obstacle avoidance which is the foundational function of advanced application has great importance. In multi-agent navigation and obstacle avoidance tasks, the decision-making interactions and dynamic changes of agents are difficult for traditional route planning algorithms or reinforcement learning algorithms with the increased complexity of the environment. The classical multi-agent reinforcement learning algorithm, Multi-agent deep deterministic policy gradient(MADDPG), solved precedent algorithms' problems of having unstationary training process and unable to deal with environment randomness. However, MADDPG ignored the temporal message hidden beneath agents' interaction with the environment. Besides, due to its CTDE technique which let each agent's critic network to calculate over all agents' action and the whole environment information, it lacks ability to scale to larger amount of agents. To deal with MADDPG's ignorance of the temporal information of the data, this article proposes a new algorithm called MADDPG-LSTMactor, which combines MADDPG with Long short term memory (LSTM). By using agent's observations of continuous timesteps as the input of its policy network, it allows the LSTM layer to process the hidden temporal message. Experimental result demonstrated that this algorithm had better performance in scenarios where the amount of agents is small. Besides, to solve MADDPG's drawback of not being efficient in scenarios where agents are too many, this article puts forward a light-weight MADDPG (MADDPG-L) algorithm, which simplifies the input of critic network. The result of experiments showed that this algorithm had better performance than MADDPG when the amount of agents was large.

Recent years have seen an increasing amount of work on embodied AI agents that can perform tasks by following human language instructions. However, most of these agents are reactive, meaning that they simply learn and imitate behaviors encountered in the training data. These reactive agents are insufficient for long-horizon complex tasks. To address this limitation, we propose a neuro-symbolic deliberative agent that, while following language instructions, proactively applies reasoning and planning based on its neural and symbolic representations acquired from past experience (e.g., natural language and egocentric vision). We show that our deliberative agent achieves greater than 70% improvement over reactive baselines on the challenging TEACh benchmark. Moreover, the underlying reasoning and planning processes, together with our modular framework, offer impressive transparency and explainability to the behaviors of the agent. This enables an in-depth understanding of the agent's capabilities, which shed light on challenges and opportunities for future embodied agents for instruction following. The code is available at https://github.com/sled-group/DANLI.