Scalable Multi-Objective Robot Reinforcement Learning through Gradient Conflict Resolution

Reinforcement Learning (RL) robot controllers usually aggregate many task objectives into one scalar reward. While large-scale proximal policy optimisation (PPO) has enabled impressive results such as robust robot locomotion in the real world, many tasks still require careful reward tuning and are brittle to local optima. Tuning cost and sub-optimality grow with the number of objectives, limiting scalability. Modelling reward vectors and their trade-offs can address these issues; however, multi-objective methods remain underused in RL for robotics because of computational cost and optimisation difficulty. In this work, we investigate the conflict between gradient contributions for each objective that emerge from scalarising the task objectives. In particular, we explicitly address the conflict between task-based rewards and terms that regularise the policy towards realistic behaviour. We propose GCR-PPO, a modification to actor-critic optimisation that decomposes the actor update into objective-wise gradients using a multi-headed critic and resolves conflicts based on the objective priority. Our methodology, GCR-PPO, is evaluated on the well-known IsaacLab manipulation and locomotion benchmarks and additional multi-objective modifications on two related tasks. We show superior scalability compared to parallel PPO (p = 0.04), without significant computational overhead. We also show higher performance with more conflicting tasks. GCR-PPO improves on large-scale PPO with an average improvement of 9.5%, with high-conflict tasks observing a greater improvement. The code is available at https://github.com/humphreymunn/GCR-PPO.

On Entropy Control in LLM-RL Algorithms

For RL algorithms, appropriate entropy control is crucial to their effectiveness. To control the policy entropy, a commonly used method is entropy regularization, which is adopted in various popular RL algorithms including PPO, SAC and A3C. Although entropy regularization proves effective in robotic and games RL conventionally, studies found that it gives weak to no gains in LLM-RL training. In this work, we study the issues of entropy bonus in LLM-RL setting. Specifically, we first argue that the conventional entropy regularization suffers from the LLM's extremely large response space and the sparsity of the optimal outputs. As a remedy, we propose AEnt, an entropy control method that utilizes a new clamped entropy bonus with an automatically adjusted coefficient. The clamped entropy is evaluated with the re-normalized policy defined on certain smaller token space, which encourages exploration within a more compact response set. In addition, the algorithm automatically adjusts entropy coefficient according to the clamped entropy value, effectively controlling the entropy-induced bias while leveraging the entropy's benefits. AEnt is tested in math-reasoning tasks under different base models and datasets, and it is observed that AEnt outperforms the baselines consistently across multiple benchmarks.

Learning on the Fly: Rapid Policy Adaptation via Differentiable Simulation

Learning control policies in simulation enables rapid, safe, and cost-effective development of advanced robotic capabilities. However, transferring these policies to the real world remains difficult due to the sim-to-real gap, where unmodeled dynamics and environmental disturbances can degrade policy performance. Existing approaches, such as domain randomization and Real2Sim2Real pipelines, can improve policy robustness, but either struggle under out-of-distribution conditions or require costly offline retraining. In this work, we approach these problems from a different perspective. Instead of relying on diverse training conditions before deployment, we focus on rapidly adapting the learned policy in the real world in an online fashion. To achieve this, we propose a novel online adaptive learning framework that unifies residual dynamics learning with real-time policy adaptation inside a differentiable simulation. Starting from a simple dynamics model, our framework refines the model continuously with real-world data to capture unmodeled effects and disturbances such as payload changes and wind. The refined dynamics model is embedded in a differentiable simulation framework, enabling gradient backpropagation through the dynamics and thus rapid, sample-efficient policy updates beyond the reach of classical RL methods like PPO. All components of our system are designed for rapid adaptation, enabling the policy to adjust to unseen disturbances within 5 seconds of training. We validate the approach on agile quadrotor control under various disturbances in both simulation and the real world. Our framework reduces hovering error by up to 81% compared to L1-MPC and 55% compared to DATT, while also demonstrating robustness in vision-based control without explicit state estimation.

Palpation Alters Auditory Pain Expressions with Gender-Specific Variations in Robopatients

Diagnostic errors remain a major cause of preventable deaths, particularly in resource-limited regions. Medical training simulators, including robopatients, play a vital role in reducing these errors by mimicking real patients for procedural training such as palpation. However, generating multimodal feedback, especially auditory pain expressions, remains challenging due to the complex relationship between palpation behavior and sound. The high-dimensional nature of pain sounds makes exploration challenging with conventional methods. This study introduces a novel experimental paradigm for pain expressivity in robopatients where they dynamically generate auditory pain expressions in response to palpation force, by co-optimizing human feedback using machine learning. Using Proximal Policy Optimization (PPO), a reinforcement learning (RL) technique optimized for continuous adaptation, our robot iteratively refines pain sounds based on real-time human feedback. This robot initializes randomized pain responses to palpation forces, and the RL agent learns to adjust these sounds to align with human preferences. The results demonstrated that the system adapts to an individual's palpation forces and sound preferences and captures a broad spectrum of pain intensity, from mild discomfort to acute distress, through RL-guided exploration of the auditory pain space. The study further showed that pain sound perception exhibits saturation at lower forces with gender specific thresholds. These findings highlight the system's potential to enhance abdominal palpation training by offering a controllable and immersive simulation platform.

Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion

Generalizing locomotion policies across diverse legged robots with varying morphologies is a key challenge due to differences in observation/action dimensions and system dynamics. In this work, we propose Multi-Loco, a novel unified framework combining a morphology-agnostic generative diffusion model with a lightweight residual policy optimized via reinforcement learning (RL). The diffusion model captures morphology-invariant locomotion patterns from diverse cross-embodiment datasets, improving generalization and robustness. The residual policy is shared across all embodiments and refines the actions generated by the diffusion model, enhancing task-aware performance and robustness for real-world deployment. We evaluated our method with a rich library of four legged robots in both simulation and real-world experiments. Compared to a standard RL framework with PPO, our approach -- replacing the Gaussian policy with a diffusion model and residual term -- achieves a 10.35% average return improvement, with gains up to 13.57% in wheeled-biped locomotion tasks. These results highlight the benefits of cross-embodiment data and composite generative architectures in learning robust, generalized locomotion skills.

PPO-BR: Dual-Signal Entropy-Reward Adaptation for Trust Region Policy Optimization

Despite Proximal Policy Optimization (PPO) dominating policy gradient methods -- from robotic control to game AI -- its static trust region forces a brittle trade-off: aggressive clipping stifles early exploration, while late-stage updates destabilize convergence. PPO-BR establishes a new paradigm in adaptive RL by fusing exploration and convergence signals into a single bounded trust region -- a theoretically grounded innovation that outperforms five SOTA baselines with less than 2% overhead. This work bridges a critical gap in phase-aware learning, enabling real-world deployment in safety-critical systems like robotic surgery within a single adaptive mechanism. PPO-BR achieves 29.1% faster convergence by combining: (1) entropy-driven expansion (epsilon up) for exploration in high-uncertainty states, and (2) reward-guided contraction (epsilon down) for convergence stability. On six diverse benchmarks (MuJoCo, Atari, sparse-reward), PPO-BR achieves 29.1% faster convergence (p < 0.001), 2.3x lower reward variance than PPO, and less than 1.8% runtime overhead with only five lines of code change. PPO-BR's simplicity and theoretical guarantees make it ready-to-deploy in safety-critical domains -- from surgical robotics to autonomous drones. In contrast to recent methods such as Group Relative Policy Optimization (GRPO), PPO-BR offers a unified entropy-reward mechanism applicable to both language models and general reinforcement learning environments.

Reduce Computational Cost In Deep Reinforcement Learning Via Randomized Policy Learning

Recent advancements in reinforcement learning (RL) have leveraged neural networks to achieve state-of-the-art performance across various control tasks. However, these successes often come at the cost of significant computational resources, as training deep neural networks requires substantial time and data. In this paper, we introduce an actor-critic algorithm that utilizes randomized neural networks to drastically reduce computational costs while maintaining strong performance. Despite its simple architecture, our method effectively solves a range of control problems, including the locomotion control of a highly dynamic 12-motor quadruped robot, and achieves results comparable to leading algorithms such as Proximal Policy Optimization (PPO). Notably, our approach does not outperform other algorithms in terms of sample efficnency but rather in terms of wall-clock training time. That is, although our algorithm requires more timesteps to converge to an optimal policy, the actual time required for training turns out to be lower.

McARL:Morphology-Control-Aware Reinforcement Learning for Generalizable Quadrupedal Locomotion

We present Morphology-Control-Aware Reinforcement Learning (McARL), a new approach to overcome challenges of hyperparameter tuning and transfer loss, enabling generalizable locomotion across robot morphologies. We use a morphology-conditioned policy by incorporating a randomized morphology vector, sampled from a defined morphology range, into both the actor and critic networks. This allows the policy to learn parameters that generalize to robots with similar characteristics. We demonstrate that a single policy trained on a Unitree Go1 robot using McARL can be transferred to a different morphology (e.g., Unitree Go2 robot) and can achieve zero-shot transfer velocity of up to 3.5 m/s without retraining or fine-tuning. Moreover, it achieves 6.0 m/s on the training Go1 robot and generalizes to other morphologies like A1 and Mini Cheetah. We also analyze the impact of morphology distance on transfer performance and highlight McARL's advantages over prior approaches. McARL achieves 44-150% higher transfer performance on Go2, Mini Cheetah, and A1 compared to PPO variants.

Fine-tuning Diffusion Policies with Backpropagation Through Diffusion Timesteps

Diffusion policies, widely adopted in decision-making scenarios such as robotics, gaming and autonomous driving, are capable of learning diverse skills from demonstration data due to their high representation power. However, the sub-optimal and limited coverage of demonstration data could lead to diffusion policies that generate sub-optimal trajectories and even catastrophic failures. While reinforcement learning (RL)-based fine-tuning has emerged as a promising solution to address these limitations, existing approaches struggle to effectively adapt Proximal Policy Optimization (PPO) to diffusion models. This challenge stems from the computational intractability of action likelihood estimation during the denoising process, which leads to complicated optimization objectives. In our experiments starting from randomly initialized policies, we find that online tuning of Diffusion Policies demonstrates much lower sample efficiency compared to directly applying PPO on MLP policies (MLP+PPO). To address these challenges, we introduce NCDPO, a novel framework that reformulates Diffusion Policy as a noise-conditioned deterministic policy. By treating each denoising step as a differentiable transformation conditioned on pre-sampled noise, NCDPO enables tractable likelihood evaluation and gradient backpropagation through all diffusion timesteps. Our experiments demonstrate that NCDPO achieves sample efficiency comparable to MLP+PPO when training from scratch, outperforming existing methods in both sample efficiency and final performance across diverse benchmarks, including continuous robot control and multi-agent game scenarios. Furthermore, our experimental results show that our method is robust to the number denoising timesteps in the Diffusion Policy.

GenPO: Generative Diffusion Models Meet On-Policy Reinforcement Learning

Recent advances in reinforcement learning (RL) have demonstrated the powerful exploration capabilities and multimodality of generative diffusion-based policies. While substantial progress has been made in offline RL and off-policy RL settings, integrating diffusion policies into on-policy frameworks like PPO remains underexplored. This gap is particularly significant given the widespread use of large-scale parallel GPU-accelerated simulators, such as IsaacLab, which are optimized for on-policy RL algorithms and enable rapid training of complex robotic tasks. A key challenge lies in computing state-action log-likelihoods under diffusion policies, which is straightforward for Gaussian policies but intractable for flow-based models due to irreversible forward-reverse processes and discretization errors (e.g., Euler-Maruyama approximations). To bridge this gap, we propose GenPO, a generative policy optimization framework that leverages exact diffusion inversion to construct invertible action mappings. GenPO introduces a novel doubled dummy action mechanism that enables invertibility via alternating updates, resolving log-likelihood computation barriers. Furthermore, we also use the action log-likelihood for unbiased entropy and KL divergence estimation, enabling KL-adaptive learning rates and entropy regularization in on-policy updates. Extensive experiments on eight IsaacLab benchmarks, including legged locomotion (Ant, Humanoid, Anymal-D, Unitree H1, Go2), dexterous manipulation (Shadow Hand), aerial control (Quadcopter), and robotic arm tasks (Franka), demonstrate GenPO's superiority over existing RL baselines. Notably, GenPO is the first method to successfully integrate diffusion policies into on-policy RL, unlocking their potential for large-scale parallelized training and real-world robotic deployment.