An emergency responder management (ERM) system dispatches responders, such as ambulances, when it receives requests for medical aid. ERM systems can also proactively reposition responders between predesignated waiting locations to cover any gaps that arise due to the prior dispatch of responders or significant changes in the distribution of anticipated requests. Optimal repositioning is computationally challenging due to the exponential number of ways to allocate responders between locations and the uncertainty in future requests. The state-of-the-art approach in proactive repositioning is a hierarchical approach based on spatial decomposition and online Monte Carlo tree search, which may require minutes of computation for each decision in a domain where seconds can save lives. We address the issue of long decision times by introducing a novel reinforcement learning (RL) approach, based on the same hierarchical decomposition, but replacing online search with learning. To address the computational challenges posed by large, variable-dimensional, and discrete state and action spaces, we propose: (1) actor-critic based agents that incorporate transformers to handle variable-dimensional states and actions, (2) projections to fixed-dimensional observations to handle complex states, and (3) combinatorial techniques to map continuous actions to discrete allocations. We evaluate our approach using real-world data from two U.S. cities, Nashville, TN and Seattle, WA. Our experiments show that compared to the state of the art, our approach reduces computation time per decision by three orders of magnitude, while also slightly reducing average ambulance response time by 5 seconds.

While the conditional sequence modeling with the transformer architecture has demonstrated its effectiveness in dealing with offline reinforcement learning (RL) tasks, it is struggle to handle out-of-distribution states and actions. Existing work attempts to address this issue by data augmentation with the learned policy or adding extra constraints with the value-based RL algorithm. However, these studies still fail to overcome the following challenges: (1) insufficiently utilizing the historical temporal information among inter-steps, (2) overlooking the local intrastep relationships among states, actions and return-to-gos (RTGs), (3) overfitting suboptimal trajectories with noisy labels. To address these challenges, we propose Decision Mamba (DM), a novel multi-grained state space model (SSM) with a self-evolving policy learning strategy. DM explicitly models the historical hidden state to extract the temporal information by using the mamba architecture. To capture the relationship among state-action-RTG triplets, a fine-grained SSM module is designed and integrated into the original coarse-grained SSM in mamba, resulting in a novel mamba architecture tailored for offline RL. Finally, to mitigate the overfitting issue on noisy trajectories, a self-evolving policy is proposed by using progressive regularization. The policy evolves by using its own past knowledge to refine the suboptimal actions, thus enhancing its robustness on noisy demonstrations. Extensive experiments on various tasks show that DM outperforms other baselines substantially.

In unsupervised environment design, reinforcement learning agents are trained on environment configurations (levels) generated by an adversary that maximises some objective. Regret is a commonly used objective that theoretically results in a minimax regret (MMR) policy with desirable robustness guarantees; in particular, the agent's maximum regret is bounded. However, once the agent reaches this regret bound on all levels, the adversary will only sample levels where regret cannot be further reduced. Although there are possible performance improvements to be made outside of these regret-maximising levels, learning stagnates. In this work, we introduce Bayesian level-perfect MMR (BLP), a refinement of the minimax regret objective that overcomes this limitation. We formally show that solving for this objective results in a subset of MMR policies, and that BLP policies act consistently with a Perfect Bayesian policy over all levels. We further introduce an algorithm, ReMiDi, that results in a BLP policy at convergence. We empirically demonstrate that training on levels from a minimax regret adversary causes learning to prematurely stagnate, but that ReMiDi continues learning.

Imitation learning provides an efficient way to teach robots dexterous skills; however, learning complex skills robustly and generalizablely usually consumes large amounts of human demonstrations. To tackle this challenging problem, we present 3D Diffusion Policy (DP3), a novel visual imitation learning approach that incorporates the power of 3D visual representations into diffusion policies, a class of conditional action generative models. The core design of DP3 is the utilization of a compact 3D visual representation, extracted from sparse point clouds with an efficient point encoder. In our experiments involving 72 simulation tasks, DP3 successfully handles most tasks with just 10 demonstrations and surpasses baselines with a 24.2% relative improvement. In 4 real robot tasks, DP3 demonstrates precise control with a high success rate of 85%, given only 40 demonstrations of each task, and shows excellent generalization abilities in diverse aspects, including space, viewpoint, appearance, and instance. Interestingly, in real robot experiments, DP3 rarely violates safety requirements, in contrast to baseline methods which frequently do, necessitating human intervention. Our extensive evaluation highlights the critical importance of 3D representations in real-world robot learning. Videos, code, and data are available on https://3d-diffusion-policy.github.io .

Meta-Reinforcement Learning (Meta-RL) agents can struggle to operate across tasks with varying environmental features that require different optimal skills (i.e., different modes of behaviours). Using context encoders based on contrastive learning to enhance the generalisability of Meta-RL agents is now widely studied but faces challenges such as the requirement for a large sample size, also referred to as the $\log$-$K$ curse. To improve RL generalisation to different tasks, we first introduce Skill-aware Mutual Information (SaMI), an optimisation objective that aids in distinguishing context embeddings according to skills, thereby equipping RL agents with the ability to identify and execute different skills across tasks. We then propose Skill-aware Noise Contrastive Estimation (SaNCE), a $K$-sample estimator used to optimise the SaMI objective. We provide a framework for equipping an RL agent with SaNCE in practice and conduct experimental validation on modified MuJoCo and Panda-gym benchmarks. We empirically find that RL agents that learn by maximising SaMI achieve substantially improved zero-shot generalisation to unseen tasks. Additionally, the context encoder equipped with SaNCE demonstrates greater robustness to reductions in the number of available samples, thus possessing the potential to overcome the $\log$-$K$ curse.

In Extreme Q-learning (XQL), Gumbel Regression is performed with an assumed Gumbel distribution for the error distribution. This allows learning of the value function without sampling out-of-distribution actions and has shown excellent performance mainly in Offline RL. However, issues remained, including the exponential term in the loss function causing instability and the potential for an error distribution diverging from the Gumbel distribution. Therefore, we propose Maclaurin Expanded Extreme Q-learning to enhance stability. In this method, applying Maclaurin expansion to the loss function in XQL enhances stability against large errors. It also allows adjusting the error distribution assumption from normal to Gumbel based on the expansion order. Our method significantly stabilizes learning in Online RL tasks from DM Control, where XQL was previously unstable. Additionally, it improves performance in several Offline RL tasks from D4RL, where XQL already showed excellent results.

This survey presents an overview of methods for learning from video (LfV) in the context of reinforcement learning (RL) and robotics. We focus on methods capable of scaling to large internet video datasets and, in the process, extracting foundational knowledge about the world's dynamics and physical human behaviour. Such methods hold great promise for developing general-purpose robots. We open with an overview of fundamental concepts relevant to the LfV-for-robotics setting. This includes a discussion of the exciting benefits LfV methods can offer (e.g., improved generalization beyond the available robot data) and commentary on key LfV challenges (e.g., missing information in video and LfV distribution shifts). Our literature review begins with an analysis of video foundation model techniques that can extract knowledge from large, heterogeneous video datasets. Next, we review methods that specifically leverage video data for robot learning. Here, we categorise work according to which RL knowledge modality (KM) benefits from the use of video data. We additionally highlight techniques for mitigating LfV challenges, including reviewing action representations that address missing action labels in video. Finally, we examine LfV datasets and benchmarks, before concluding with a discussion of challenges and opportunities in LfV. Here, we advocate for scalable foundation model approaches that can leverage the full range of internet video data, and that target the learning of the most promising RL KMs: the policy and dynamics model. Overall, we hope this survey will serve as a comprehensive reference for the emerging field of LfV, catalysing further research in the area and facilitating progress towards the development of general-purpose robots.

Imitation learning enables autonomous agents to learn from human examples, without the need for a reward signal. Still, if the provided dataset does not encapsulate the task correctly, or when the task is too complex to be modeled, such agents fail to reproduce the expert policy. We propose to recover from these failures through online adaptation. Our approach combines the action proposal coming from a pre-trained policy with relevant experience recorded by an expert. The combination results in an adapted action that closely follows the expert. Our experiments show that an adapted agent performs better than its pure imitation learning counterpart. Notably, adapted agents can achieve reasonable performance even when the base, non-adapted policy catastrophically fails.

Sim-to-real transfer presents a difficult challenge, where models trained in simulation are to be deployed in the real world. The distribution shift between the two settings leads to biased representations of the perceived real-world environment, and thus to suboptimal predictions. In this work, we tackle the challenge of sim-to-real transfer of reinforcement learning (RL) agents for coverage path planning (CPP). In CPP, the task is for a robot to find a path that visits every point of a confined area. Specifically, we consider the case where the environment is unknown, and the agent needs to plan the path online while mapping the environment. We bridge the sim-to-real gap through a semi-virtual environment with a simulated sensor and obstacles, while including real robot kinematics and real-time aspects. We investigate what level of fine-tuning is needed for adapting to a realistic setting, comparing to an agent trained solely in simulation. We find that a high model inference frequency is sufficient for reducing the sim-to-real gap, while fine-tuning degrades performance initially. By training the model in simulation and deploying it at a high inference frequency, we transfer state-of-the-art results from simulation to the real domain, where direct learning would take in the order of weeks with manual interaction, i.e., would be completely infeasible.

Ensuring the safety of Reinforcement Learning (RL) is crucial for its deployment in real-world applications. Nevertheless, managing the trade-off between reward and safety during exploration presents a significant challenge. Improving reward performance through policy adjustments may adversely affect safety performance. In this study, we aim to address this conflicting relation by leveraging the theory of gradient manipulation. Initially, we analyze the conflict between reward and safety gradients. Subsequently, we tackle the balance between reward and safety optimization by proposing a soft switching policy optimization method, for which we provide convergence analysis. Based on our theoretical examination, we provide a safe RL framework to overcome the aforementioned challenge, and we develop a Safety-MuJoCo Benchmark to assess the performance of safe RL algorithms. Finally, we evaluate the effectiveness of our method on the Safety-MuJoCo Benchmark and a popular safe RL benchmark, Omnisafe. Experimental results demonstrate that our algorithms outperform several state-of-the-art baselines in terms of balancing reward and safety optimization.