Swarm intelligence emerges from decentralised interactions among simple agents, enabling collective problem-solving. This study establishes a theoretical equivalence between pheromone-mediated aggregation in \celeg\ and reinforcement learning (RL), demonstrating how stigmergic signals function as distributed reward mechanisms. We model engineered nematode swarms performing foraging tasks, showing that pheromone dynamics mathematically mirror cross-learning updates, a fundamental RL algorithm. Experimental validation with data from literature confirms that our model accurately replicates empirical \celeg\ foraging patterns under static conditions. In dynamic environments, persistent pheromone trails create positive feedback loops that hinder adaptation by locking swarms into obsolete choices. Through computational experiments in multi-armed bandit scenarios, we reveal that introducing a minority of exploratory agents insensitive to pheromones restores collective plasticity, enabling rapid task switching. This behavioural heterogeneity balances exploration-exploitation trade-offs, implementing swarm-level extinction of outdated strategies. Our results demonstrate that stigmergic systems inherently encode distributed RL processes, where environmental signals act as external memory for collective credit assignment. By bridging synthetic biology with swarm robotics, this work advances programmable living systems capable of resilient decision-making in volatile environments.

Penetration testing, the simulation of cyberattacks to identify security vulnerabilities, presents a sequential decision-making problem well-suited for reinforcement learning (RL) automation. Like many applications of RL to real-world problems, partial observability presents a major challenge, as it invalidates the Markov property present in Markov Decision Processes (MDPs). Partially Observable MDPs require history aggregation or belief state estimation to learn successful policies. We investigate stochastic, partially observable penetration testing scenarios over host networks of varying size, aiming to better reflect real-world complexity through more challenging and representative benchmarks. This approach leads to the development of more robust and transferable policies, which are crucial for ensuring reliable performance across diverse and unpredictable real-world environments. Using vanilla Proximal Policy Optimization (PPO) as a baseline, we compare a selection of PPO variants designed to mitigate partial observability, including frame-stacking, augmenting observations with historical information, and employing recurrent or transformer-based architectures. We conduct a systematic empirical analysis of these algorithms across different host network sizes. We find that this task greatly benefits from history aggregation. Converging three times faster than other approaches. Manual inspection of the learned policies by the algorithms reveals clear distinctions and provides insights that go beyond quantitative results.

We propose a probabilistic shared-control solution for navigation, called Robot Trajectron V2 (RT-V2), that enables accurate intent prediction and safe, effective assistance in human-robot interaction. RT-V2 jointly models a user's long-term behavioral patterns and their noisy, low-dimensional control signals by combining a prior intent model with a posterior update that accounts for real-time user input and environmental context. The prior captures the multimodal and history-dependent nature of user intent using recurrent neural networks and conditional variational autoencoders, while the posterior integrates this with uncertain user commands to infer desired actions. We conduct extensive experiments to validate RT-V2 across synthetic benchmarks, human-computer interaction studies with keyboard input, and brain-machine interface experiments with non-human primates. Results show that RT-V2 outperforms the state of the art in intent estimation, provides safe and efficient navigation support, and adequately balances user autonomy with assistive intervention. By unifying probabilistic modeling, reinforcement learning, and safe optimization, RT-V2 offers a principled and generalizable approach to shared control for diverse assistive technologies.

Exploration in reinforcement learning (RL) remains challenging, particularly in sparse-reward settings. While foundation models possess strong semantic priors, their capabilities as zero-shot exploration agents in classic RL benchmarks are not well understood. We benchmark LLMs and VLMs on multi-armed bandits, Gridworlds, and sparse-reward Atari to test zero-shot exploration. Our investigation reveals a key limitation: while VLMs can infer high-level objectives from visual input, they consistently fail at precise low-level control: the "knowing-doing gap". To analyze a potential bridge for this gap, we investigate a simple on-policy hybrid framework in a controlled, best-case scenario. Our results in this idealized setting show that VLM guidance can significantly improve early-stage sample efficiency, providing a clear analysis of the potential and constraints of using foundation models to guide exploration rather than for end-to-end control.

Achieving diverse and stable dexterous grasping for general and deformable objects remains a fundamental challenge in robotics, due to high-dimensional action spaces and uncertainty in perception. In this paper, we present D3Grasp, a multimodal perception-guided reinforcement learning framework designed to enable Diverse and Deformable Dexterous Grasping. We firstly introduce a unified multimodal representation that integrates visual and tactile perception to robustly grasp common objects with diverse properties. Second, we propose an asymmetric reinforcement learning architecture that exploits privileged information during training while preserving deployment realism, enhancing both generalization and sample efficiency. Third, we meticulously design a training strategy to synthesize contact-rich, penetration-free, and kinematically feasible grasps with enhanced adaptability to deformable and contact-sensitive objects. Extensive evaluations confirm that D3Grasp delivers highly robust performance across large-scale and diverse object categories, and substantially advances the state of the art in dexterous grasping for deformable and compliant objects, even under perceptual uncertainty and real-world disturbances. D3Grasp achieves an average success rate of 95.1% in real-world trials,outperforming prior methods on both rigid and deformable objects benchmarks.

Policy-based reinforcement learning currently plays an important role in improving LLMs on mathematical reasoning tasks. However, existing rollout-based reinforcement learning methods (GRPO, DAPO, GSPO, etc.) fail to explicitly consider LLMs' learning ability for samples of different difficulty levels, which is contrary to the human cognitive process of mathematical reasoning tasks from easy to difficult. Intuitively, we find that the variance of the rollout group's reward in RLVR partly reflects the difficulty of the current sample for LLMs. Samples that are too easy or too difficult have a lower variance, while samples with moderate difficulty have a higher variance. Based on this, we propose VCRL, a curriculum reinforcement learning framework that dynamically controls the difficulty of training samples based on the variance of group rewards. Experiments on five mathematical benchmarks and two models reveal the advantages of VCRL over the current LLM RL baselines.

Diffusion-based world models have demonstrated strong capabilities in synthesizing realistic long-horizon trajectories for offline reinforcement learning (RL). However, many existing methods do not directly generate actions alongside states and rewards, limiting their compatibility with standard value-based offline RL algorithms that rely on one-step temporal difference (TD) learning. While prior work has explored joint modeling of states, rewards, and actions to address this issue, such formulations often lead to increased training complexity and reduced performance in practice. We propose \textbf{DAWM}, a diffusion-based world model that generates future state-reward trajectories conditioned on the current state, action, and return-to-go, paired with an inverse dynamics model (IDM) for efficient action inference. This modular design produces complete synthetic transitions suitable for one-step TD-based offline RL, enabling effective and computationally efficient training. Empirically, we show that conservative offline RL algorithms such as TD3BC and IQL benefit significantly from training on these augmented trajectories, consistently outperforming prior diffusion-based baselines across multiple tasks in the D4RL benchmark.

Multi-Agent Reinforcement Learning (MARL) is a growing research area which gained significant traction in recent years, extending Deep RL applications to a much wider range of problems. A particularly challenging class of problems in this domain is Heterogeneous Multi-Agent Reinforcement Learning (HeMARL), where agents with different sensors, resources, or capabilities must cooperate based on local information. The large number of real-world situations involving heterogeneous agents makes it an attractive research area, yet underexplored, as most MARL research focuses on homogeneous agents (e.g., a swarm of identical robots). In MARL and single-agent RL, standardized environments such as ALE and SMAC have allowed to establish recognized benchmarks to measure progress. However, there is a clear lack of such standardized testbed for cooperative HeMARL. As a result, new research in this field often uses simple environments, where most algorithms perform near optimally, or uses weakly heterogeneous MARL environments.

Diffusion probability models have shown significant promise in offline reinforcement learning by directly modeling trajectory sequences. However, existing approaches primarily focus on time-domain features while overlooking frequency-domain features, leading to frequency shift and degraded performance according to our observation. In this paper, we investigate the RL problem from a new perspective of the frequency domain. We first observe that time-domain-only approaches inadvertently introduce shifts in the low-frequency components of the frequency domain, which results in trajectory instability and degraded performance. To address this issue, we propose Wavelet Fourier Diffuser (WFDiffuser), a novel diffusion-based RL framework that integrates Discrete Wavelet Transform to decompose trajectories into low- and high-frequency components. To further enhance diffusion modeling for each component, WFDiffuser employs Short-Time Fourier Transform and cross attention mechanisms to extract frequency-domain features and facilitate cross-frequency interaction. Extensive experiment results on the D4RL benchmark demonstrate that WFDiffuser effectively mitigates frequency shift, leading to smoother, more stable trajectories and improved decision-making performance over existing methods.

Graphical User Interface (GUI) agents have demonstrated remarkable progress in automating complex user interface interactions through reinforcement learning. However, current approaches face a fundamental dilemma: offline RL enables stable training on pre-collected trajectories, but struggles with multi-step task execution for lack of trajectory-level reward signals; online RL captures these signals through environment interaction, but suffers from sparse rewards and prohibitive deployment costs. To address it, we present Semi-online Reinforcement Learning, a novel paradigm that simulates online RL on offline trajectories. During each rollout process, we preserve the original model output within the multi-turn dialogue, where a Patch Module adaptively recovers the divergence between rollout and expert trajectories. To capture long-term training signals, Semi-online RL introduces discounted future returns into the reward computation and optimizes the policy with weighted step-level and episode-level advantages. We further introduce Semi-Online Performance (SOP), a metric that aligns better with true online performance, serving as a practical and effective proxy for real-world evaluation. Experiments show that ours UI-S1-7B achieves SOT A performance among 7B models across four dynamic benchmarks, with significant gains over the base model (e.g., +12.0% on AndroidWorld, +23.8% on AITW), demonstrating significant progress in bridging the gap between of-fline training efficiency and online multi-turn reasoning. Our proposed Semi-online RL simulates online RL on offline static trajectories, which enhances multi-turn agent capabilities more efficiently. Work done during internship at Tongyi Lab, Alibaba Group.