A universal controller for any robot morphology would greatly improve computational and data efficiency. Steps have been made towards such multi-robot control by utilizing contextual information about the properties of individual robots and exploiting their modular structure in the architecture of deep reinforcement learning agents. When the robots have highly dissimilar morphologies, however, this becomes a challenging problem, especially when the agent must generalize to new, unseen robots. In this paper, we consider the fact that contextual features are often only partially available, but that they can be recovered through modular interactions. This can allow for better multi-robot control and generalization to contexts that are not seen during training. To this extent, we implement a transformer-based architecture with shared modular recurrence and evaluate its (generalization) performance on a large set of MuJoCo robots. The results show a substantial improvement in zero-shot generalization performance on robots with unseen dynamics, kinematics, and topologies, in four different environments.
Conventional reinforcement learning (RL) ap proaches often struggle to learn effective policies under sparse reward conditions, necessitating the manual design of complex, task-specific reward functions. To address this limitation, rein forcement learning from human feedback (RLHF) has emerged as a promising strategy that complements hand-crafted rewards with human-derived evaluation signals. However, most existing RLHF methods depend on explicit feedback mechanisms such as button presses or preference labels, which disrupt the natural interaction process and impose a substantial cognitive load on the user. We propose a novel reinforcement learning from implicit human feedback (RLIHF) framework that utilizes non-invasive electroencephalography (EEG) signals, specifically error-related potentials (ErrPs), to provide continuous, implicit feedback without requiring explicit user intervention. The proposed method adopts a pre-trained decoder to transform raw EEG signals into probabilistic reward components, en abling effective policy learning even in the presence of sparse external rewards. We evaluate our approach in a simulation environment built on the MuJoCo physics engine, using a Kinova Gen2 robotic arm to perform a complex pick-and-place task that requires avoiding obstacles while manipulating target objects. The results show that agents trained with decoded EEG feedback achieve performance comparable to those trained with dense, manually designed rewards. These findings validate the potential of using implicit neural feedback for scalable and human-aligned reinforcement learning in interactive robotics.
Deep reinforcement learning (DRL) agents excel in solving complex decision-making tasks across various domains. However, they often require a substantial number of training steps and a vast experience replay buffer, leading to significant computational and resource demands. To address these challenges, we introduce a novel theoretical result that leverages the Neyman-Rubin potential outcomes framework into DRL. Unlike most methods that focus on bounding the counterfactual loss, we establish a causal bound on the factual loss, which is analogous to the on-policy loss in DRL. This bound is computed by storing past value network outputs in the experience replay buffer, effectively utilizing data that is usually discarded. Extensive experiments across the Atari 2600 and MuJoCo domains on various agents, such as DQN and SAC, achieve up to 2,427% higher reward ratio, outperforming the same agents without our proposed term, and reducing the experience replay buffer size by up to 96%, significantly improving sample efficiency at negligible cost.
Achieving fast and stable off-policy learning in deep reinforcement learning (RL) is challenging. Most existing methods rely on semi-gradient temporal-difference (TD) methods for their simplicity and efficiency, but are consequently susceptible to divergence. While more principled approaches like Gradient TD (GTD) methods have strong convergence guarantees, they have rarely been used in deep RL. Recent work introduced the Generalized Projected Bellman Error ($\GPBE$), enabling GTD methods to work efficiently with nonlinear function approximation. However, this work is only limited to one-step methods, which are slow at credit assignment and require a large number of samples. In this paper, we extend the $\GPBE$ objective to support multistep credit assignment based on the $\lambda$-return and derive three gradient-based methods that optimize this new objective. We provide both a forward-view formulation compatible with experience replay and a backward-view formulation compatible with streaming algorithms. Finally, we evaluate the proposed algorithms and show that they outperform both PPO and StreamQ in MuJoCo and MinAtar environments, respectively. Code available at https://github.com/esraaelelimy/gtd\_algos
Reward hacking in Reinforcement Learning (RL) systems poses a critical threat to the deployment of autonomous agents, where agents exploit flaws in reward functions to achieve high scores without fulfilling intended objectives. Despite growing awareness of this problem, systematic detection and mitigation approaches remain limited. This paper presents a large-scale empirical study of reward hacking across diverse RL environments and algorithms. We analyze 15,247 training episodes across 15 RL environments (Atari, MuJoCo, custom domains) and 5 algorithms (PPO, SAC, DQN, A3C, Rainbow), implementing automated detection algorithms for six categories of reward hacking: specification gaming, reward tampering, proxy optimization, objective misalignment, exploitation patterns, and wireheading. Our detection framework achieves 78.4% precision and 81.7% recall across environments, with computational overhead under 5%. Through controlled experiments varying reward function properties, we demonstrate that reward density and alignment with true objectives significantly impact hacking frequency ($p < 0.001$, Cohen's $d = 1.24$). We validate our approach through three simulated application studies representing recommendation systems, competitive gaming, and robotic control scenarios. Our mitigation techniques reduce hacking frequency by up to 54.6% in controlled scenarios, though we find these trade-offs are more challenging in practice due to concept drift, false positive costs, and adversarial adaptation. All detection algorithms, datasets, and experimental protocols are publicly available to support reproducible research in RL safety.
Deploying reinforcement learning (RL) policies in real-world involves significant challenges, including distribution shifts, safety concerns, and the impracticality of direct interactions during policy refinement. Existing methods, such as domain randomization (DR) and off-dynamics RL, enhance policy robustness by direct interaction with the target domain, an inherently unsafe practice. We propose Uncertainty-Aware RL (UARL), a novel framework that prioritizes safety during training by addressing Out-Of-Distribution (OOD) detection and policy adaptation without requiring direct interactions in target domain. UARL employs an ensemble of critics to quantify policy uncertainty and incorporates progressive environmental randomization to prepare the policy for diverse real-world conditions. By iteratively refining over high-uncertainty regions of the state space in simulated environments, UARL enhances robust generalization to the target domain without explicitly training on it. We evaluate UARL on MuJoCo benchmarks and a quadrupedal robot, demonstrating its effectiveness in reliable OOD detection, improved performance, and enhanced sample efficiency compared to baselines.
Transfer learning in Reinforcement Learning (RL) enables agents to leverage knowledge from source tasks to accelerate learning in target tasks. While prior work, such as the Attend, Adapt, and Transfer (A2T) framework, addresses negative transfer and selective transfer, other critical challenges remain underexplored. This paper introduces the Generalized Adaptive Transfer Network (GATN), a deep RL architecture designed to tackle task generalization across domains, robustness to environmental changes, and computational efficiency in transfer. GATN employs a domain-agnostic representation module, a robustness-aware policy adapter, and an efficient transfer scheduler to achieve these goals. We evaluate GATN on diverse benchmarks, including Atari 2600, MuJoCo, and a custom chatbot dialogue environment, demonstrating superior performance in cross-domain generalization, resilience to dynamic environments, and reduced computational overhead compared to baselines. Our findings suggest GATN is a versatile framework for real-world RL applications, such as adaptive chatbots and robotic control.
Meta-reinforcement learning requires utilizing prior task distribution information obtained during exploration to rapidly adapt to unknown tasks. The efficiency of an agent's exploration hinges on accurately identifying the current task. Recent Bayes-Adaptive Deep RL approaches often rely on reconstructing the environment's reward signal, which is challenging in sparse reward settings, leading to suboptimal exploitation. Inspired by bisimulation metrics, which robustly extracts behavioral similarity in continuous MDPs, we propose SimBelief-a novel meta-RL framework via measuring similarity of task belief in Bayes-Adaptive MDP (BAMDP). SimBelief effectively extracts common features of similar task distributions, enabling efficient task identification and exploration in sparse reward environments. We introduce latent task belief metric to learn the common structure of similar tasks and incorporate it into the specific task belief. By learning the latent dynamics across task distributions, we connect shared latent task belief features with specific task features, facilitating rapid task identification and adaptation. Our method outperforms state-of-the-art baselines on sparse reward MuJoCo and panda-gym tasks.
rQdia regularizes Q-value distributions with augmented images in pixel-based deep reinforcement learning. With a simple auxiliary loss, that equalizes these distributions via MSE, rQdia boosts DrQ and SAC on 9/12 and 10/12 tasks respectively in the MuJoCo Continuous Control Suite from pixels, and Data-Efficient Rainbow on 18/26 Atari Arcade environments. Gains are measured in both sample efficiency and longer-term training. Moreover, the addition of rQdia finally propels model-free continuous control from pixels over the state encoding baseline.
Bias problems in the estimation of $Q$-values are a well-known obstacle that slows down convergence of $Q$-learning and actor-critic methods. One of the reasons of the success of modern RL algorithms is partially a direct or indirect overestimation reduction mechanism. We propose an easy to implement method built on top of distributional reinforcement learning (DRL) algorithms to deal with the overestimation in a locally adaptive way. Our framework is simple to implement, existing distributional algorithms can be improved with a few lines of code. We provide theoretical evidence and use double $Q$-learning to show how to include locally adaptive overestimation control in existing algorithms. Experiments are provided for tabular, Atari, and MuJoCo environments.
Off-policy deep reinforcement learning (RL) typically leverages replay buffers for reusing past experiences during learning. This can help improve sample efficiency when the collected data is informative and aligned with the learning objectives; when that is not the case, it can have the effect of "polluting" the replay buffer with data which can exacerbate optimization challenges in addition to wasting environment interactions due to wasteful sampling. We argue that sampling these uninformative and wasteful transitions can be avoided by addressing the sunk cost fallacy, which, in the context of deep RL, is the tendency towards continuing an episode until termination. To address this, we propose learn to stop (LEAST), a lightweight mechanism that enables strategic early episode termination based on Q-value and gradient statistics, which helps agents recognize when to terminate unproductive episodes early. We demonstrate that our method improves learning efficiency on a variety of RL algorithms, evaluated on both the MuJoCo and DeepMind Control Suite benchmarks.
Imitation learning is a central problem in reinforcement learning where the goal is to learn a policy that mimics the expert's behavior. In practice, it is often challenging to learn the expert policy from a limited number of demonstrations accurately due to the complexity of the state space. Moreover, it is essential to explore the environment and collect data to achieve beyond-expert performance. To overcome these challenges, we propose a novel imitation learning algorithm called Imitation Learning with Double Exploration (ILDE), which implements exploration in two aspects: (1) optimistic policy optimization via an exploration bonus that rewards state-action pairs with high uncertainty to potentially improve the convergence to the expert policy, and (2) curiosity-driven exploration of the states that deviate from the demonstration trajectories to potentially yield beyond-expert performance. Empirically, we demonstrate that ILDE outperforms the state-of-the-art imitation learning algorithms in terms of sample efficiency and achieves beyond-expert performance on Atari and MuJoCo tasks with fewer demonstrations than in previous work. We also provide a theoretical justification of ILDE as an uncertainty-regularized policy optimization method with optimistic exploration, leading to a regret growing sublinearly in the number of episodes.
Contact forces pose a major challenge for gradient-based optimization of robot dynamics as they introduce jumps in the system's velocities. Penalty-based simulators, such as MuJoCo, simplify gradient computation by softening the contact forces. However, realistically simulating hard contacts requires very stiff contact settings, which leads to incorrect gradients when using automatic differentiation. On the other hand, using non-stiff settings strongly increases the sim-to-real gap. We analyze the contact computation of penalty-based simulators to identify the causes of gradient errors. Then, we propose DiffMJX, which combines adaptive integration with MuJoCo XLA, to notably improve gradient quality in the presence of hard contacts. Finally, we address a key limitation of contact gradients: they vanish when objects do not touch. To overcome this, we introduce Contacts From Distance (CFD), a mechanism that enables the simulator to generate informative contact gradients even before objects are in contact. To preserve physical realism, we apply CFD only in the backward pass using a straight-through trick, allowing us to compute useful gradients without modifying the forward simulation.
Deep off-policy actor-critic algorithms have emerged as the leading framework for reinforcement learning in continuous control domains. However, most of these algorithms suffer from poor sample efficiency, especially in environments with sparse rewards. In this paper, we take a step towards addressing this issue by providing a principled directed exploration strategy. We propose Wasserstein Barycenter Soft Actor-Critic (WBSAC) algorithm, which benefits from a pessimistic actor for temporal difference learning and an optimistic actor to promote exploration. This is achieved by using the Wasserstein barycenter of the pessimistic and optimistic policies as the exploration policy and adjusting the degree of exploration throughout the learning process. We compare WBSAC with state-of-the-art off-policy actor-critic algorithms and show that WBSAC is more sample-efficient on MuJoCo continuous control tasks.
Although diffusion models have achieved strong results in decision-making tasks, their slow inference speed remains a key limitation. While the consistency model offers a potential solution, its applications to decision-making often struggle with suboptimal demonstrations or rely on complex concurrent training of multiple networks. In this work, we propose a novel approach to consistency distillation for offline reinforcement learning that directly incorporates reward optimization into the distillation process. Our method enables single-step generation while maintaining higher performance and simpler training. Empirical evaluations on the Gym MuJoCo benchmarks and long horizon planning demonstrate that our approach can achieve an 8.7% improvement over previous state-of-the-art while offering up to 142x speedup over diffusion counterparts in inference time.
We study the off-dynamics offline reinforcement learning problem, where the goal is to learn a policy from offline datasets collected from source and target domains with mismatched transition. Existing off-dynamics offline RL methods typically either filter source transitions that resemble those of the target domain or apply reward augmentation to source data, both constrained by the limited transitions available from the target domain. As a result, the learned policy is unable to explore target domain beyond the offline datasets. We propose MOBODY, a Model-Based Off-Dynamics offline RL algorithm that addresses this limitation by enabling exploration of the target domain via learned dynamics. MOBODY generates new synthetic transitions in the target domain through model rollouts, which are used as data augmentation during offline policy learning. Unlike existing model-based methods that learn dynamics from a single domain, MOBODY tackles the challenge of mismatched dynamics by leveraging both source and target datasets. Directly merging these datasets can bias the learned model toward source dynamics. Instead, MOBODY learns target dynamics by discovering a shared latent representation of states and transitions across domains through representation learning. To stabilize training, MOBODY incorporates a behavior cloning loss that regularizes the policy. Specifically, we introduce a Q-weighted behavior cloning loss that regularizes the policy toward actions with high target-domain Q-values, rather than uniformly imitating all actions in the dataset. These Q-values are learned from an enhanced target dataset composed of offline target data, augmented source data, and rollout data from the learned target dynamics. We evaluate MOBODY on MuJoCo benchmarks and show that it significantly outperforms state-of-the-art baselines, with especially pronounced improvements in challenging scenarios.
We propose a method that enables large language models (LLMs) to control embodied agents by directly mapping continuous observation vectors to continuous action vectors. Initially, the LLMs generate a control strategy based on a textual description of the agent, its environment, and the intended goal. This strategy is then iteratively refined through a learning process in which the LLMs are repeatedly prompted to improve the current strategy, using performance feedback and sensory-motor data collected during its evaluation. The method is validated on classic control tasks from the Gymnasium library and the inverted pendulum task from the MuJoCo library. In most cases, it successfully identifies optimal or high-performing solutions by integrating symbolic knowledge derived through reasoning with sub-symbolic sensory-motor data gathered as the agent interacts with its environment.
Deep reinforcement learning (RL) agents frequently suffer from neuronal activity loss, which impairs their ability to adapt to new data and learn continually. A common method to quantify and address this issue is the tau-dormant neuron ratio, which uses activation statistics to measure the expressive ability of neurons. While effective for simple MLP-based agents, this approach loses statistical power in more complex architectures. To address this, we argue that in advanced RL agents, maintaining a neuron's learning capacity, its ability to adapt via gradient updates, is more critical than preserving its expressive ability. Based on this insight, we shift the statistical objective from activations to gradients, and introduce GraMa (Gradient Magnitude Neural Activity Metric), a lightweight, architecture-agnostic metric for quantifying neuron-level learning capacity. We show that GraMa effectively reveals persistent neuron inactivity across diverse architectures, including residual networks, diffusion models, and agents with varied activation functions. Moreover, resetting neurons guided by GraMa (ReGraMa) consistently improves learning performance across multiple deep RL algorithms and benchmarks, such as MuJoCo and the DeepMind Control Suite.
Real-world datasets collected from sensors or human inputs are prone to noise and errors, posing significant challenges for applying offline reinforcement learning (RL). While existing methods have made progress in addressing corrupted actions and rewards, they remain insufficient for handling corruption in high-dimensional state spaces and for cases where multiple elements in the dataset are corrupted simultaneously. Diffusion models, known for their strong denoising capabilities, offer a promising direction for this problem-but their tendency to overfit noisy samples limits their direct applicability. To overcome this, we propose Ambient Diffusion-Guided Dataset Recovery (ADG), a novel approach that pioneers the use of diffusion models to tackle data corruption in offline RL. First, we introduce Ambient Denoising Diffusion Probabilistic Models (DDPM) from approximated distributions, which enable learning on partially corrupted datasets with theoretical guarantees. Second, we use the noise-prediction property of Ambient DDPM to distinguish between clean and corrupted data, and then use the clean subset to train a standard DDPM. Third, we employ the trained standard DDPM to refine the previously identified corrupted data, enhancing data quality for subsequent offline RL training. A notable strength of ADG is its versatility-it can be seamlessly integrated with any offline RL algorithm. Experiments on a range of benchmarks, including MuJoCo, Kitchen, and Adroit, demonstrate that ADG effectively mitigates the impact of corrupted data and improves the robustness of offline RL under various noise settings, achieving state-of-the-art results.
Due to their expressive capacity, diffusion models have shown great promise in offline RL and imitation learning. Diffusion Actor-Critic with Entropy Regulator (DACER) extended this capability to online RL by using the reverse diffusion process as a policy approximator, trained end-to-end with policy gradient methods, achieving strong performance. However, this comes at the cost of requiring many diffusion steps, which significantly hampers training efficiency, while directly reducing the steps leads to noticeable performance degradation. Critically, the lack of inference efficiency becomes a significant bottleneck for applying diffusion policies in real-time online RL settings. To improve training and inference efficiency while maintaining or even enhancing performance, we propose a Q-gradient field objective as an auxiliary optimization target to guide the denoising process at each diffusion step. Nonetheless, we observe that the independence of the Q-gradient field from the diffusion time step negatively impacts the performance of the diffusion policy. To address this, we introduce a temporal weighting mechanism that enables the model to efficiently eliminate large-scale noise in the early stages and refine actions in the later stages. Experimental results on MuJoCo benchmarks and several multimodal tasks demonstrate that the DACER2 algorithm achieves state-of-the-art performance in most MuJoCo control tasks with only five diffusion steps, while also exhibiting stronger multimodality compared to DACER.