Language models (LMs) are trained on collections of documents, written by individual human agents to achieve specific goals in an outside world. During training, LMs have access only to text of these documents, with no direct evidence of the internal states of the agents that produced them -- a fact often used to argue that LMs are incapable of modeling goal-directed aspects of human language production and comprehension. Can LMs trained on text learn anything at all about the relationship between language and use? I argue that LMs are models of intentional communication in a specific, narrow sense. When performing next word prediction given a textual context, an LM can infer and represent properties of an agent likely to have produced that context. These representations can in turn influence subsequent LM generation in the same way that agents' communicative intentions influence their language. I survey findings from the recent literature showing that -- even in today's non-robust and error-prone models -- LMs infer and use representations of fine-grained communicative intentions and more abstract beliefs and goals. Despite the limited nature of their training data, they can thus serve as building blocks for systems that communicate and act intentionally.

Secondly, the communication improves its policy iteratively by learning from observations delay can interfere with the cooperation between agents to achieve a given goal. RL, with a single agent to decide by introducing delays in action-making (Chen et al. 2021) the behavior of all entities, faces various challenges, such and uncertainty on the arrival time of information. Previous as scalability (Yan et al. 2021) and privacy issues (Yuan, work (Kim et al. 2019) prevents endless waiting by setting Chung, and Fu 2022). To this end, the extension from singleagent a predefined and constant bound for the waiting time, but it RL to multi-agent RL (MARL) (Hernandez-Leal, Kartal, may restrain potential cooperation if it is set too short and and Taylor 2019) is favorable. MARL (Hernandez-Leal, conversely may cause meaningless waiting. Therefore, such Kartal, and Taylor 2019) has been widely used in various a constant timer is inflexible and cannot be adapted to the tasks, such as real-time resource allocation (Yuan et al. dynamics in the communication networks.

Endowing nonlinear systems with safe behavior is increasingly important in modern control. This task is particularly challenging for real-life control systems that must operate safely in dynamically changing environments. This paper develops a framework for safety-critical control in dynamic environments, by establishing the notion of environmental control barrier functions (ECBFs). The framework is able to guarantee safety even in the presence of input delay, by accounting for the evolution of the environment during the delayed response of the system. The underlying control synthesis relies on predicting the future state of the system and the environment over the delay interval, with robust safety guarantees against prediction errors. The efficacy of the proposed method is demonstrated by a simple adaptive cruise control problem and a more complex robotics application on a Segway platform.

Complex, multi-objective missions require the coordination of heterogeneous robots at multiple inter-connected levels, such as coalition formation, scheduling, and motion planning. This challenge is exacerbated by dynamic changes, such as sensor and actuator failures, communication loss, and unexpected delays. We introduce Dynamic Iterative Task Allocation Graph Search (D-ITAGS) to \textit{simultaneously} address coalition formation, scheduling, and motion planning in dynamic settings involving heterogeneous teams. D-ITAGS achieves resilience via two key characteristics: i) interleaved execution, and ii) targeted repair. \textit{Interleaved execution} enables an effective search for solutions at each layer while avoiding incompatibility with other layers. \textit{Targeted repair} identifies and repairs parts of the existing solution impacted by a given disruption, while conserving the rest. In addition to algorithmic contributions, we provide theoretical insights into the inherent trade-off between time and resource optimality in these settings and derive meaningful bounds on schedule suboptimality. Our experiments reveal that i) D-ITAGS is significantly faster than recomputation from scratch in dynamic settings, with little to no loss in solution quality, and ii) the theoretical suboptimality bounds consistently hold in practice.

Sentient or not, MineDojo got Nvidia's researchers the NeurIPS 2022 Outstanding Datasets and Benchmarks Paper Award, demonstrating their belief that large language models in the future will be "embodied agents that proactively take actions, endlessly explore the world, and continuously self-improve." The current version of this AI agent learned Minecraft's flexible gameplay using a massive online database of more than 7,000 wiki pages, millions of Reddit threads and 300,000 hours of recorded gameplay.

It's been well over a decade since I finished a Cognitive Architectures course at MIT (9.364) under the late professor Whitman Richards. My final project was a little thing called "Agent Collaboration Using Anigrafs." Anigrafs were a pedagogical cognitive architecture that Richards defined. I wrote some code to implement Anigrafs and hooked it to simulated robots, which I called "Anibots." What follows is essentially my 2008 final report to Prof. Richards. I noticed several years later that he published Anigrafs: Experiments in Cooperative Cognitive Architecture as a book from MIT Press. TLDR: Identical robots can cooperate if they use a type of mental network that votes (with the Condorcet method) on what behavior to do next. The development goal is to achieve collaboration of situated agents to perform shared tasks and/or goals.

In this work, we study the problem of Embodied Referring Expression Grounding, where an agent needs to navigate in a previously unseen environment and localize a remote object described by a concise high-level natural language instruction. When facing such a situation, a human tends to imagine what the destination may look like and to explore the environment based on prior knowledge of the environmental layout, such as the fact that a bathroom is more likely to be found near a bedroom than a kitchen. We have designed an autonomous agent called Layout-aware Dreamer (LAD), including two novel modules, that is, the Layout Learner and the Goal Dreamer to mimic this cognitive decision process. The Layout Learner learns to infer the room category distribution of neighboring unexplored areas along the path for coarse layout estimation, which effectively introduces layout common sense of room-to-room transitions to our agent. To learn an effective exploration of the environment, the Goal Dreamer imagines the destination beforehand. Our agent achieves new state-of-the-art performance on the public leaderboard of the REVERIE dataset in challenging unseen test environments with improvement in navigation success (SR) by 4.02% and remote grounding success (RGS) by 3.43% compared to the previous state-of-the-art. The code is released at https://github.com/zehao-wang/LAD

Multi-agent reinforcement learning (MARL) suffers from the non-stationarity problem, which is the ever-changing targets at every iteration when multiple agents update their policies at the same time. Starting from first principle, in this paper, we manage to solve the non-stationarity problem by proposing bidirectional action-dependent Q-learning (ACE). Central to the development of ACE is the sequential decision-making process wherein only one agent is allowed to take action at one time. Within this process, each agent maximizes its value function given the actions taken by the preceding agents at the inference stage. In the learning phase, each agent minimizes the TD error that is dependent on how the subsequent agents have reacted to their chosen action. Given the design of bidirectional dependency, ACE effectively turns a multiagent MDP into a single-agent MDP. We implement the ACE framework by identifying the proper network representation to formulate the action dependency, so that the sequential decision process is computed implicitly in one forward pass. To validate ACE, we compare it with strong baselines on two MARL benchmarks. Empirical experiments demonstrate that ACE outperforms the state-of-the-art algorithms on Google Research Football and StarCraft Multi-Agent Challenge by a large margin. In particular, on SMAC tasks, ACE achieves 100% success rate on almost all the hard and super-hard maps. We further study extensive research problems regarding ACE, including extension, generalization, and practicability. Code is made available to facilitate further research.

We consider the problem of autonomous channel access (AutoCA), where a group of terminals tries to discover a communication strategy with an access point (AP) via a common wireless channel in a distributed fashion. Due to the irregular topology and the limited communication range of terminals, a practical challenge for AutoCA is the hidden terminal problem, which is notorious in wireless networks for deteriorating the throughput and delay performances. To meet the challenge, this paper presents a new multi-agent deep reinforcement learning paradigm, dubbed MADRL-HT, tailored for AutoCA in the presence of hidden terminals. MADRL-HT exploits topological insights and transforms the observation space of each terminal into a scalable form independent of the number of terminals. To compensate for the partial observability, we put forth a look-back mechanism such that the terminals can infer behaviors of their hidden terminals from the carrier sensed channel states as well as feedback from the AP. A window-based global reward function is proposed, whereby the terminals are instructed to maximize the system throughput while balancing the terminals' transmission opportunities over the course of learning. Extensive numerical experiments verified the superior performance of our solution benchmarked against the legacy carrier-sense multiple access with collision avoidance (CSMA/CA) protocol.

This paper studies a distributed policy gradient in collaborative multi-agent reinforcement learning (MARL), where agents over a communication network aim to find the optimal policy to maximize the average of all agents' local returns. Due to the non-concave performance function of policy gradient, the existing distributed stochastic optimization methods for convex problems cannot be directly used for policy gradient in MARL. This paper proposes a distributed policy gradient with variance reduction and gradient tracking to address the high variances of policy gradient, and utilizes importance weight to solve the {distribution shift} problem in the sampling process. We then provide an upper bound on the mean-squared stationary gap, which depends on the number of iterations, the mini-batch size, the epoch size, the problem parameters, and the network topology. We further establish the sample and communication complexity to obtain an $\epsilon$-approximate stationary point. Numerical experiments are performed to validate the effectiveness of the proposed algorithm.