Scaling Off-Policy Reinforcement Learning with Batch and Weight Normalization

Reinforcement learning has achieved significant milestones, but sample efficiency remains a bottleneck for real-world applications. Recently, CrossQ has demonstrated state-of-the-art sample efficiency with a low update-to-data (UTD) ratio of 1. In this work, we explore CrossQ's scaling behavior with higher UTD ratios. We identify challenges in the training dynamics, which are emphasized by higher UTD ratios. To address these, we integrate weight normalization into the CrossQ framework, a solution that stabilizes training, has been shown to prevent potential loss of plasticity and keeps the effective learning rate constant. Our proposed approach reliably scales with increasing UTD ratios, achieving competitive performance across 25 challenging continuous control tasks on the DeepMind Control Suite and Myosuite benchmarks, notably the complex dog and humanoid environments. This work eliminates the need for drastic interventions, such as network resets, and offers a simple yet robust pathway for improving sample efficiency and scalability in model-free reinforcement learning.

Stable Port-Hamiltonian Neural Networks

In recent years, nonlinear dynamic system identification using artificial neural networks has garnered attention due to its manifold potential applications in virtually all branches of science and engineering. However, purely data-driven approaches often struggle with extrapolation and may yield physically implausible forecasts. Furthermore, the learned dynamics can exhibit instabilities, making it difficult to apply such models safely and robustly. This article proposes stable port-Hamiltonian neural networks, a machine learning architecture that incorporates the physical biases of energy conservation or dissipation while guaranteeing global Lyapunov stability of the learned dynamics. Evaluations with illustrative examples and real-world measurement data demonstrate the model's ability to generalize from sparse data, outperforming purely data-driven approaches and avoiding instability issues. In addition, the model's potential for data-driven surrogate modeling is highlighted in application to multi-physics simulation data.

DIME:Diffusion-Based Maximum Entropy Reinforcement Learning

Maximum entropy reinforcement learning (MaxEnt-RL) has become the standard approach to RL due to its beneficial exploration properties. Traditionally, policies are parameterized using Gaussian distributions, which significantly limits their representational capacity. Diffusion-based policies offer a more expressive alternative, yet integrating them into MaxEnt-RL poses challenges--primarily due to the intractability of computing their marginal entropy. To overcome this, we propose Diffusion-Based Maximum Entropy RL (DIME). DIME leverages recent advances in approximate inference with diffusion models to derive a lower bound on the maximum entropy objective. Additionally, we propose a policy iteration scheme that provably converges to the optimal diffusion policy. Our method enables the use of expressive diffusion-based policies while retaining the principled exploration benefits of MaxEnt-RL, significantly outperforming other diffusion-based methods on challenging high-dimensional control benchmarks. It is also competitive with state-of-the-art non-diffusion based RL methods while requiring fewer algorithmic design choices and smaller update-to-data ratios, reducing computational complexity.

Noise-conditioned Energy-based Annealed Rewards (NEAR): A Generative Framework for Imitation Learning from Observation

This paper introduces a new imitation learning framework based on energy-based generative models capable of learning complex, physics-dependent, robot motion policies through state-only expert motion trajectories. Our algorithm, called Noise-conditioned Energy-based Annealed Rewards (NEAR), constructs several perturbed versions of the expert's motion data distribution and learns smooth, and well-defined representations of the data distribution's energy function using denoising score matching. We propose to use these learnt energy functions as reward functions to learn imitation policies via reinforcement learning. We also present a strategy to gradually switch between the learnt energy functions, ensuring that the learnt rewards are always well-defined in the manifold of policy-generated samples. We evaluate our algorithm on complex humanoid tasks such as locomotion and martial arts and compare it with state-only adversarial imitation learning algorithms like Adversarial Motion Priors (AMP). Our framework sidesteps the optimisation challenges of adversarial imitation learning techniques and produces results comparable to AMP in several quantitative metrics across multiple imitation settings.

Fast and Robust Visuomotor Riemannian Flow Matching Policy

Diffusion-based visuomotor policies excel at learning complex robotic tasks by effectively combining visual data with high-dimensional, multi-modal action distributions. However, diffusion models often suffer from slow inference due to costly denoising processes or require complex sequential training arising from recent distilling approaches. This paper introduces Riemannian Flow Matching Policy (RFMP), a model that inherits the easy training and fast inference capabilities of flow matching (FM). Moreover, RFMP inherently incorporates geometric constraints commonly found in realistic robotic applications, as the robot state resides on a Riemannian manifold. To enhance the robustness of RFMP, we propose Stable RFMP (SRFMP), which leverages LaSalle's invariance principle to equip the dynamics of FM with stability to the support of a target Riemannian distribution. Rigorous evaluation on eight simulated and real-world tasks show that RFMP successfully learns and synthesizes complex sensorimotor policies on Euclidean and Riemannian spaces with efficient training and inference phases, outperforming Diffusion Policies while remaining competitive with Consistency Policies.

Grasp Diffusion Network: Learning Grasp Generators from Partial Point Clouds with Diffusion Models in SO(3)xR3

Grasping objects successfully from a single-view camera is crucial in many robot manipulation tasks. An approach to solve this problem is to leverage simulation to create large datasets of pairs of objects and grasp poses, and then learn a conditional generative model that can be prompted quickly during deployment. However, the grasp pose data is highly multimodal since there are several ways to grasp an object. Hence, in this work, we learn a grasp generative model with diffusion models to sample candidate grasp poses given a partial point cloud of an object. A novel aspect of our method is to consider diffusion in the manifold space of rotations and to propose a collision-avoidance cost guidance to improve the grasp success rate during inference. To accelerate grasp sampling we use recent techniques from the diffusion literature to achieve faster inference times. We show in simulation and real-world experiments that our approach can grasp several objects from raw depth images with $90\%$ success rate and benchmark it against several baselines.

Particle-based 6D Object Pose Estimation from Point Clouds using Diffusion Models

Object pose estimation from a single view remains a challenging problem. In particular, partial observability, occlusions, and object symmetries eventually result in pose ambiguity. To account for this multimodality, this work proposes training a diffusion-based generative model for 6D object pose estimation. During inference, the trained generative model allows for sampling multiple particles, i.e., pose hypotheses. To distill this information into a single pose estimate, we propose two novel and effective pose selection strategies that do not require any additional training or computationally intensive operations. Moreover, while many existing methods for pose estimation primarily focus on the image domain and only incorporate depth information for final pose refinement, our model solely operates on point cloud data. The model thereby leverages recent advancements in point cloud processing and operates upon an SE(3)-equivariant latent space that forms the basis for the particle selection strategies and allows for improved inference times. Our thorough experimental results demonstrate the competitive performance of our approach on the Linemod dataset and showcase the effectiveness of our design choices. Code is available at https://github.com/zitronian/6DPoseDiffusion .

Global Tensor Motion Planning

Batch planning is increasingly crucial for the scalability of robotics tasks and dataset generation diversity. This paper presents Global Tensor Motion Planning (GTMP) -- a sampling-based motion planning algorithm comprising only tensor operations. We introduce a novel discretization structure represented as a random multipartite graph, enabling efficient vectorized sampling, collision checking, and search. We provide an early theoretical investigation showing that GTMP exhibits probabilistic completeness while supporting modern GPU/TPU. Additionally, by incorporating smooth structures into the multipartite graph, GTMP directly plans smooth splines without requiring gradient-based optimization. Experiments on lidar-scanned occupancy maps and the MotionBenchMarker dataset demonstrate GTMP's computation efficiency in batch planning compared to baselines, underscoring GTMP's potential as a robust, scalable planner for diverse applications and large-scale robot learning tasks.

A Retrospective on the Robot Air Hockey Challenge: Benchmarking Robust, Reliable, and Safe Learning Techniques for Real-world Robotics

Machine learning methods have a groundbreaking impact in many application domains, but their application on real robotic platforms is still limited. Despite the many challenges associated with combining machine learning technology with robotics, robot learning remains one of the most promising directions for enhancing the capabilities of robots. When deploying learning-based approaches on real robots, extra effort is required to address the challenges posed by various real-world factors. To investigate the key factors influencing real-world deployment and to encourage original solutions from different researchers, we organized the Robot Air Hockey Challenge at the NeurIPS 2023 conference. We selected the air hockey task as a benchmark, encompassing low-level robotics problems and high-level tactics. Different from other machine learning-centric benchmarks, participants need to tackle practical challenges in robotics, such as the sim-to-real gap, low-level control issues, safety problems, real-time requirements, and the limited availability of real-world data. Furthermore, we focus on a dynamic environment, removing the typical assumption of quasi-static motions of other real-world benchmarks. The competition's results show that solutions combining learning-based approaches with prior knowledge outperform those relying solely on data when real-world deployment is challenging. Our ablation study reveals which real-world factors may be overlooked when building a learning-based solution. The successful real-world air hockey deployment of best-performing agents sets the foundation for future competitions and follow-up research directions.

TacEx: GelSight Tactile Simulation in Isaac Sim -- Combining Soft-Body and Visuotactile Simulators

Training robot policies in simulation is becoming increasingly popular; nevertheless, a precise, reliable, and easy-to-use tactile simulator for contact-rich manipulation tasks is still missing. To close this gap, we develop TacEx -- a modular tactile simulation framework. We embed a state-of-the-art soft-body simulator for contacts named GIPC and vision-based tactile simulators Taxim and FOTS into Isaac Sim to achieve robust and plausible simulation of the visuotactile sensor GelSight Mini. We implement several Isaac Lab environments for Reinforcement Learning (RL) leveraging our TacEx simulation, including object pushing, lifting, and pole balancing. We validate that the simulation is stable and that the high-dimensional observations, such as the gel deformation and the RGB images from the GelSight camera, can be used for training. The code, videos, and additional results will be released online https://sites.google.com/view/tacex.