When creating new reinforcement learning tasks, practitioners often accelerate the learning process by incorporating into the task several accessory components, such as breaking the environment interaction into independent episodes and frequently resetting the environment. Although they can enable the learning of complex intelligent behaviors, such task accessories can result in unnatural task setups and hinder long-term performance in the real world. In this work, we explore the challenges of learning without episode terminations and robot embodiment resets using the Soft Actor-Critic (SAC) algorithm. To learn without terminations, we present a continuing version of the SAC algorithm and show that, with simple modifications to the reward functions of existing tasks, continuing SAC can perform as well as or better than episodic SAC while reducing the sensitivity of performance to the value of the discount rate $γ$. On a modified Gym Reacher task, we investigate possible explanations for the failure of continuing SAC when learning without embodiment resets. Our results suggest that embodiment resets help with exploration of the state space in the SAC algorithm, and removing embodiment resets can lead to poor exploration of the state space and failure of or significantly slower learning. Finally, on additional simulated tasks and a real-robot vision task, we show that increasing the entropy of the policy when performance trends worse or remains static is an effective intervention for recovering the performance lost due to not using embodiment resets.

This paper applies the authors' recent results on asynchronous stochastic approximation (SA) in the Borkar-Meyn framework to reinforcement learning in average-reward semi-Markov decision processes (SMDPs). We establish the convergence of an asynchronous SA analogue of Schweitzer's classical relative value iteration algorithm, RVI Q-learning, for finite-space, weakly communicating SMDPs. In particular, we show that the algorithm converges almost surely to a compact, connected subset of solutions to the average-reward optimality equation, with convergence to a unique, sample path-dependent solution under additional stepsize and asynchrony conditions. Moreover, to make full use of the SA framework, we introduce new monotonicity conditions for estimating the optimal reward rate in RVI Q-learning. These conditions substantially expand the previously considered algorithmic framework and are addressed through novel arguments in the stability and convergence analysis of RVI Q-learning.

How much does a trained RL policy actually use its past observations? We propose \emph{Temporal Range}, a model-agnostic metric that treats first-order sensitivities of multiple vector outputs across a temporal window to the input sequence as a temporal influence profile and summarizes it by the magnitude-weighted average lag. Temporal Range is computed via reverse-mode automatic differentiation from the Jacobian blocks $\partial y_s/\partial x_t\in\mathbb{R}^{c\times d}$ averaged over final timesteps $s\in\{t+1,\dots,T\}$ and is well-characterized in the linear setting by a small set of natural axioms. Across diagnostic and control tasks (POPGym; flicker/occlusion; Copy-$k$) and architectures (MLPs, RNNs, SSMs), Temporal Range (i) remains small in fully observed control, (ii) scales with the task's ground-truth lag in Copy-$k$, and (iii) aligns with the minimum history window required for near-optimal return as confirmed by window ablations. We also report Temporal Range for a compact Long Expressive Memory (LEM) policy trained on the task, using it as a proxy readout of task-level memory. Our axiomatic treatment draws on recent work on range measures, specialized here to temporal lag and extended to vector-valued outputs in the RL setting. Temporal Range thus offers a practical per-sequence readout of memory dependence for comparing agents and environments and for selecting the shortest sufficient context.

Artificial intelligence methods are increasingly being explored for managing wildfires and other natural hazards. In particular, reinforcement learning (RL) is a promising path towards improving outcomes in such uncertain decision-making scenarios and moving beyond reactive strategies. However, training RL agents requires many environment interactions, and the speed of existing wildfire simulators is a severely limiting factor. We introduce $\texttt{JaxWildfire}$, a simulator underpinned by a principled probabilistic fire spread model based on cellular automata. It is implemented in JAX and enables vectorized simulations using $\texttt{vmap}$, allowing high throughput of simulations on GPUs. We demonstrate that $\texttt{JaxWildfire}$ achieves 6-35x speedup over existing software and enables gradient-based optimization of simulator parameters. Furthermore, we show that $\texttt{JaxWildfire}$ can be used to train RL agents to learn wildfire suppression policies. Our work is an important step towards enabling the advancement of RL techniques for managing natural hazards.

This paper introduces a framework that integrates reinforcement learning (RL) with autonomous agents to enable continuous improvement in the automated process of software test cases authoring from business requirement documents within Quality Engineering (QE) workflows. Conventional systems employing Large Language Models (LLMs) generate test cases from static knowledge bases, which fundamentally limits their capacity to enhance performance over time. Our proposed Reinforcement Infused Agentic RAG (Retrieve, Augment, Generate) framework overcomes this limitation by employing AI agents that learn from QE feedback, assessments, and defect discovery outcomes to automatically improve their test case generation strategies. The system combines specialized agents with a hybrid vector-graph knowledge base that stores and retrieves software testing knowledge. Through advanced RL algorithms, specifically Proximal Policy Optimization (PPO) and Deep Q-Networks (DQN), these agents optimize their behavior based on QE-reported test effectiveness, defect detection rates, and workflow metrics. As QEs execute AI-generated test cases and provide feedback, the system learns from this expert guidance to improve future iterations. Experimental validation on enterprise Apple projects yielded substantive improvements: a 2.4% increase in test generation accuracy (from 94.8% to 97.2%), and a 10.8% improvement in defect detection rates. The framework establishes a continuous knowledge refinement loop driven by QE expertise, resulting in progressively superior test case quality that enhances, rather than replaces, human testing capabilities.

Resource allocation remains NP-hard due to combinatorial complexity. While deep reinforcement learning (DRL) methods, such as the Rainbow Deep Q-Network (DQN), improve scalability through prioritized replay and distributional heads, classical function approximators limit their representational power. We introduce Variational Quantum Rainbow DQN (VQR-DQN), which integrates ring-topology variational quantum circuits with Rainbow DQN to leverage quantum superposition and entanglement. We frame the human resource allocation problem (HRAP) as a Markov decision process (MDP) with combinatorial action spaces based on officer capabilities, event schedules, and transition times. On four HRAP benchmarks, VQR-DQN achieves 26.8% normalized makespan reduction versus random baselines and outperforms Double DQN and classical Rainbow DQN by 4.9-13.4%. These gains align with theoretical connections between circuit expressibility, entanglement, and policy quality, demonstrating the potential of quantum-enhanced DRL for large-scale resource allocation. Our implementation is available at: https://github.com/Analytics-Everywhere-Lab/qtrl/.

We introduce a system for real-time sperm whale rendezvous at sea using an autonomous uncrewed aerial vehicle. Our system employs model-based reinforcement learning that combines in situ sensor data with an empirical whale dive model to guide navigation decisions. Key challenges include (i) real-time acoustic tracking in the presence of multiple whales, (ii) distributed communication and decision-making for robot deployments, and (iii) on-board signal processing and long-range detection from fish-trackers. We evaluate our system by conducting rendezvous with sperm whales at sea in Dominica, performing hardware experiments on land, and running simulations using whale trajectories interpolated from marine biologists' surface observations.

Abstract--This paper proposes a hierarchical trajectory planning framework for UA Vs operating under adversarial jamming conditions. Leveraging Bayesian Active Inference, the approach combines expert-generated demonstrations with probabilistic generative modeling to encode high-level symbolic planning, low-level motion policies, and wireless signal feedback. During deployment, the UA V performs online inference to anticipate interference, localize jammers, and adapt its trajectory accordingly--without prior knowledge of jammer locations. Simulation results demonstrate that the proposed method achieves near-expert performance, significantly reducing communication interference and mission cost compared to model-free reinforcement learning baselines, while maintaining robust generalization in dynamic environments. Unmanned Aerial V ehicles (UA Vs) play a crucial role in military, public, and civilian applications due to their compact size, flexible deployment capabilities, and outstanding performance.

Large language model post-training relies on reinforcement learning to improve model capability and alignment quality. However, the off-policy training paradigm introduces distribution shift, which often pushes the policy beyond the trust region, leading to training instabilities manifested as fluctuations in policy entropy and unstable gradients. Although PPO-Clip mitigates this issue through importance clipping, it still overlooks the global distributional shift of actions. To address these challenges, we propose using the entropy ratio between the current and previous policies as a new global metric that effectively quantifies the relative change in policy exploration throughout updates. Building on this metric, we introduce an \textbf{Entropy Ratio Clipping} (ERC) mechanism that imposes bidirectional constraints on the entropy ratio. This stabilizes policy updates at the global distribution level and compensates for the inability of PPO-clip to regulate probability shifts of un-sampled actions. We integrate ERC into both DAPO and GPPO reinforcement learning algorithms. Experiments across multiple benchmarks show that ERC consistently improves performance.

Background: Deep Deterministic Policy Gradient-based reinforcement learning algorithms utilize Actor-Critic architectures, where both networks are typically trained using identical batches of replayed transitions. However, the learning objectives and update dynamics of the Actor and Critic differ, raising concerns about whether uniform transition usage is optimal. Objectives: We aim to improve the performance of deep deterministic policy gradient algorithms by decoupling the transition batches used to train the Actor and the Critic. Our goal is to design an experience replay mechanism that provides appropriate learning signals to each component by using separate, tailored batches. Methods: We introduce Decoupled Prioritized Experience Replay (DPER), a novel approach that allows independent sampling of transition batches for the Actor and the Critic. DPER can be integrated into any off-policy deep reinforcement learning algorithm that operates in continuous control domains. We combine DPER with the state-of-the-art Twin Delayed DDPG algorithm and evaluate its performance across standard continuous control benchmarks. Results: DPER outperforms conventional experience replay strategies such as vanilla experience replay and prioritized experience replay in multiple MuJoCo tasks from the OpenAI Gym suite. Conclusions: Our findings show that decoupling experience replay for Actor and Critic networks can enhance training dynamics and final policy quality. DPER offers a generalizable mechanism that enhances performance for a wide class of actor-critic off-policy reinforcement learning algorithms.