Reinforcement Learning has achieved significant success in generating complex behavior but often requires extensive reward function engineering. Adversarial variants of Imitation Learning and Inverse Reinforcement Learning offer an alternative by learning policies from expert demonstrations via a discriminator. Employing discriminators increases their data- and computational efficiency over the standard approaches; however, results in sensitivity to imperfections in expert data. We propose RILe, a teacher-student system that achieves both robustness to imperfect data and efficiency. In RILe, the student learns an action policy while the teacher dynamically adjusts a reward function based on the student's performance and its alignment with expert demonstrations. By tailoring the reward function to both performance of the student and expert similarity, our system reduces dependence on the discriminator and, hence, increases robustness against data imperfections. Experiments show that RILe outperforms existing methods by 2x in settings with limited or noisy expert data.

Restless multi-armed bandits (RMABs) have been widely utilized to address resource allocation problems with Markov reward processes (MRPs). Existing works often assume that the dynamics of MRPs are known prior, which makes the RMAB problem solvable from an optimization perspective. Nevertheless, an efficient learning-based solution for RMABs with unknown system dynamics remains an open problem. In this paper, we study the cooperative resource allocation problem with unknown system dynamics of MRPs. This problem can be modeled as a multi-agent online RMAB problem, where multiple agents collaboratively learn the system dynamics while maximizing their accumulated rewards. We devise a federated online RMAB framework to mitigate the communication overhead and data privacy issue by adopting the federated learning paradigm. Based on this framework, we put forth a Federated Thompson Sampling-enabled Whittle Index (FedTSWI) algorithm to solve this multi-agent online RMAB problem. The FedTSWI algorithm enjoys a high communication and computation efficiency, and a privacy guarantee. Moreover, we derive a regret upper bound for the FedTSWI algorithm. Finally, we demonstrate the effectiveness of the proposed algorithm on the case of online multi-user multi-channel access. Numerical results show that the proposed algorithm achieves a fast convergence rate of $\mathcal{O}(\sqrt{T\log(T)})$ and better performance compared with baselines. More importantly, its sample complexity decreases with the number of agents.

Autonomous mobile robots (AMRs) have been a rapidly expanding research topic for the past decade. Unlike their counterpart, the automated guided vehicle (AGV), AMRs can make decisions and do not need any previously installed infrastructure to navigate. Recent technological developments in hardware and software have made them more feasible, especially in warehouse environments. Traditionally, most wasted warehouse expenses come from the logistics of moving material from one point to another, and is exhaustive for humans to continuously walk those distances while carrying a load. Here, AMRs can help by working with humans to cut down the time and effort of these repetitive tasks, improving performance and reducing the fatigue of their human collaborators. This literature review covers the recent developments in AMR technology including hardware, robotic control, and system control. This paper also discusses examples of current AMR producers, their robots, and the software that is used to control them. We conclude with future research topics and where we see AMRs developing in the warehouse environment.

Syntax-guided synthesis is commonly used to generate programs encoding policies. In this approach, the set of programs, that can be written in a domain-specific language defines the search space, and an algorithm searches within this space for programs that encode strong policies. In this paper, we propose an alternative method for synthesizing programmatic policies, where we search within an approximation of the language's semantic space. We hypothesized that searching in semantic spaces is more sample-efficient compared to syntax-based spaces. Our rationale is that the search is more efficient if the algorithm evaluates different agent behaviors as it searches through the space, a feature often missing in syntax-based spaces. This is because small changes in the syntax of a program often do not result in different agent behaviors. We define semantic spaces by learning a library of programs that present different agent behaviors. Then, we approximate the semantic space by defining a neighborhood function for local search algorithms, where we replace parts of the current candidate program with programs from the library. We evaluated our hypothesis in a real-time strategy game called MicroRTS. Empirical results support our hypothesis that searching in semantic spaces can be more sample-efficient than searching in syntax-based spaces.

The ability to traverse an unknown environment is crucial for autonomous robot operations. However, due to the limited sensing capabilities and system constraints, approaching this problem with a single robot agent can be slow, costly, and unsafe. For example, in planetary exploration missions, the wear on the wheels of a rover from abrasive terrain should be minimized at all costs as reparations are infeasible. On the other hand, utilizing a scouting robot such as a micro aerial vehicle (MAV) has the potential to reduce wear and time costs and increasing safety of a follower robot. This work proposes a novel cooperative IPP framework that allows a scout (e.g., an MAV) to efficiently explore the minimum-cost-path for a follower (e.g., a rover) to reach the goal. We derive theoretic guarantees for our algorithm, and prove that the algorithm always terminates, always finds the optimal path if it exists, and terminates early when the found path is shown to be optimal or infeasible. We show in thorough experimental evaluation that the guarantees hold in practice, and that our algorithm is 22.5% quicker to find the optimal path and 15% quicker to terminate compared to existing methods.

Large language model (LLM)-based mobile agents are increasingly popular due to their capability to interact directly with mobile phone Graphic User Interfaces (GUIs) and their potential to autonomously manage daily tasks. Despite their promising prospects in both academic and industrial sectors, little research has focused on benchmarking the performance of existing mobile agents, due to the inexhaustible states of apps and the vague definition of feasible action sequences. To address this challenge, we propose an efficient and user-friendly benchmark, MobileAgentBench, designed to alleviate the burden of extensive manual testing. We initially define 100 tasks across 10 open-source apps, categorized by multiple levels of difficulty. Subsequently, we evaluate several existing mobile agents, including AppAgent and MobileAgent, to thoroughly and systematically compare their performance. All materials are accessible on our project webpage: https://MobileAgentBench.github.io,

Coordination in multiplayer games enables players to avoid the lose-lose outcome that often arises at Nash equilibria. However, designing a coordination mechanism typically requires the consideration of the joint actions of all players, which becomes intractable in large-scale games. We develop a novel coordination mechanism, termed reduced rank correlated equilibria, which reduces the number of joint actions to be considered and thereby mitigates computational complexity. The idea is to approximate the set of all joint actions with the actions used in a set of pre-computed Nash equilibria via a convex hull operation. In a game with n players and each player having m actions, the proposed mechanism reduces the number of joint actions considered from O(m^n) to O(mn). We demonstrate the application of the proposed mechanism to an air traffic queue management problem. Compared with the correlated equilibrium-a popular benchmark coordination mechanism-the proposed approach is capable of solving a problem involving four thousand times more joint actions while yielding similar or better performance in terms of a fairness indicator and showing a maximum optimality gap of 0.066% in terms of the average delay cost. In the meantime, it yields a solution that shows up to 99.5% improvement in a fairness indicator and up to 50.4% reduction in average delay cost compared to the Nash solution, which does not involve coordination.

Reinforcement learning struggles in the face of long-horizon tasks and sparse goals due to the difficulty in manual reward specification. While existing methods address this by adding intrinsic rewards, they may fail to provide meaningful guidance in long-horizon decision-making tasks with large state and action spaces, lacking purposeful exploration. Inspired by human cognition, we propose a new multi-modal model-based RL approach named Dreaming with Large Language Models (DLLM). DLLM integrates the proposed hinting subgoals from the LLMs into the model rollouts to encourage goal discovery and reaching in challenging tasks. By assigning higher intrinsic rewards to samples that align with the hints outlined by the language model during model rollouts, DLLM guides the agent toward meaningful and efficient exploration. Extensive experiments demonstrate that the DLLM outperforms recent methods in various challenging, sparse-reward environments such as HomeGrid, Crafter, and Minecraft by 27.7\%, 21.1\%, and 9.9\%, respectively.

In this paper, we consider multi-objective reinforcement learning, which arises in many real-world problems with multiple optimization goals. We approach the problem with a max-min framework focusing on fairness among the multiple goals and develop a relevant theory and a practical model-free algorithm under the max-min framework. The developed theory provides a theoretical advance in multi-objective reinforcement learning, and the proposed algorithm demonstrates a notable performance improvement over existing baseline methods.

Active inference helps us simulate adaptive behavior and decision-making in biological and artificial agents. Building on our previous work exploring the relationship between active inference, well-being, resilience, and sustainability, we present a computational model of an agent learning sustainable resource management strategies in both static and dynamic environments. The agent's behavior emerges from optimizing its own well-being, represented by prior preferences, subject to beliefs about environmental dynamics. In a static environment, the agent learns to consistently consume resources to satisfy its needs. In a dynamic environment where resources deplete and replenish based on the agent's actions, the agent adapts its behavior to balance immediate needs with long-term resource availability. This demonstrates how active inference can give rise to sustainable and resilient behaviors in the face of changing environmental conditions. We discuss the implications of our model, its limitations, and suggest future directions for integrating more complex agent-environment interactions. Our work highlights active inference's potential for understanding and shaping sustainable behaviors.