Alexander Winkler

Zhengyi Luo, Jinkun Cao, Josh Merel et al. · cmu

We present a universal motion representation that encompasses a comprehensive range of motor skills for physics-based humanoid control. Due to the high dimensionality of humanoids and the inherent difficulties in reinforcement learning, prior methods have focused on learning skill embeddings for a narrow range of movement styles (e.g. locomotion, game characters) from specialized motion datasets. This limited scope hampers their applicability in complex tasks. We close this gap by significantly increasing the coverage of our motion representation space. To achieve this, we first learn a motion imitator that can imitate all of human motion from a large, unstructured motion dataset. We then create our motion representation by distilling skills directly from the imitator. This is achieved by using an encoder-decoder structure with a variational information bottleneck. Additionally, we jointly learn a prior conditioned on proprioception (humanoid's own pose and velocities) to improve model expressiveness and sampling efficiency for downstream tasks. By sampling from the prior, we can generate long, stable, and diverse human motions. Using this latent space for hierarchical RL, we show that our policies solve tasks using human-like behavior. We demonstrate the effectiveness of our motion representation by solving generative tasks (e.g. strike, terrain traversal) and motion tracking using VR controllers.

Daniele Reda, Jungdam Won, Yuting Ye et al. · eth-zurich

Avatars are important to create interactive and immersive experiences in virtual worlds. One challenge in animating these characters to mimic a user's motion is that commercial AR/VR products consist only of a headset and controllers, providing very limited sensor data of the user's pose. Another challenge is that an avatar might have a different skeleton structure than a human and the mapping between them is unclear. In this work we address both of these challenges. We introduce a method to retarget motions in real-time from sparse human sensor data to characters of various morphologies. Our method uses reinforcement learning to train a policy to control characters in a physics simulator. We only require human motion capture data for training, without relying on artist-generated animations for each avatar. This allows us to use large motion capture datasets to train general policies that can track unseen users from real and sparse data in real-time. We demonstrate the feasibility of our approach on three characters with different skeleton structure: a dinosaur, a mouse-like creature and a human. We show that the avatar poses often match the user surprisingly well, despite having no sensor information of the lower body available. We discuss and ablate the important components in our framework, specifically the kinematic retargeting step, the imitation, contact and action reward as well as our asymmetric actor-critic observations. We further explore the robustness of our method in a variety of settings including unbalancing, dancing and sports motions.

Alexander Winkler, Jungdam Won, Yuting Ye · eth-zurich

Real-time tracking of human body motion is crucial for interactive and immersive experiences in AR/VR. However, very limited sensor data about the body is available from standalone wearable devices such as HMDs (Head Mounted Devices) or AR glasses. In this work, we present a reinforcement learning framework that takes in sparse signals from an HMD and two controllers, and simulates plausible and physically valid full body motions. Using high quality full body motion as dense supervision during training, a simple policy network can learn to output appropriate torques for the character to balance, walk, and jog, while closely following the input signals. Our results demonstrate surprisingly similar leg motions to ground truth without any observations of the lower body, even when the input is only the 6D transformations of the HMD. We also show that a single policy can be robust to diverse locomotion styles, different body sizes, and novel environments.

Sunmin Lee, Sebastian Starke, Yuting Ye et al. · eth-zurich

Replicating a user's pose from only wearable sensors is important for many AR/VR applications. Most existing methods for motion tracking avoid environment interaction apart from foot-floor contact due to their complex dynamics and hard constraints. However, in daily life people regularly interact with their environment, e.g. by sitting on a couch or leaning on a desk. Using Reinforcement Learning, we show that headset and controller pose, if combined with physics simulation and environment observations can generate realistic full-body poses even in highly constrained environments. The physics simulation automatically enforces the various constraints necessary for realistic poses, instead of manually specifying them as in many kinematic approaches. These hard constraints allow us to achieve high-quality interaction motions without typical artifacts such as penetration or contact sliding. We discuss three features, the environment representation, the contact reward and scene randomization, crucial to the performance of the method. We demonstrate the generality of the approach through various examples, such as sitting on chairs, a couch and boxes, stepping over boxes, rocking a chair and turning an office chair. We believe these are some of the highest-quality results achieved for motion tracking from sparse sensor with scene interaction.

David C. Gordon, Alexander Winkler, Julian Bedei et al.

Model Predictive Control (MPC) provides an optimal control solution based on a cost function while allowing for the implementation of process constraints. As a model-based optimal control technique, the performance of MPC strongly depends on the model used where a trade-off between model computation time and prediction performance exists. One solution is the integration of MPC with a machine learning (ML) based process model which are quick to evaluate online. This work presents the experimental implementation of a deep neural network (DNN) based nonlinear MPC for Homogeneous Charge Compression Ignition (HCCI) combustion control. The DNN model consists of a Long Short-Term Memory (LSTM) network surrounded by fully connected layers which was trained using experimental engine data and showed acceptable prediction performance with under 5% error for all outputs. Using this model, the MPC is designed to track the Indicated Mean Effective Pressure (IMEP) and combustion phasing trajectories, while minimizing several parameters. Using the acados software package to enable the real-time implementation of the MPC on an ARM Cortex A72, the optimization calculations are completed within 1.4 ms. The external A72 processor is integrated with the prototyping engine controller using a UDP connection allowing for rapid experimental deployment of the NMPC. The IMEP trajectory following of the developed controller was excellent, with a root-mean-square error of 0.133 bar, in addition to observing process constraints.

Siwei Zhang, Bharat Lal Bhatnagar, Yuanlu Xu et al.

We propose RoHM, an approach for robust 3D human motion reconstruction from monocular RGB(-D) videos in the presence of noise and occlusions. Most previous approaches either train neural networks to directly regress motion in 3D or learn data-driven motion priors and combine them with optimization at test time. The former do not recover globally coherent motion and fail under occlusions; the latter are time-consuming, prone to local minima, and require manual tuning. To overcome these shortcomings, we exploit the iterative, denoising nature of diffusion models. RoHM is a novel diffusion-based motion model that, conditioned on noisy and occluded input data, reconstructs complete, plausible motions in consistent global coordinates. Given the complexity of the problem -- requiring one to address different tasks (denoising and infilling) in different solution spaces (local and global motion) -- we decompose it into two sub-tasks and learn two models, one for global trajectory and one for local motion. To capture the correlations between the two, we then introduce a novel conditioning module, combining it with an iterative inference scheme. We apply RoHM to a variety of tasks -- from motion reconstruction and denoising to spatial and temporal infilling. Extensive experiments on three popular datasets show that our method outperforms state-of-the-art approaches qualitatively and quantitatively, while being faster at test time. The code is available at https://sanweiliti.github.io/ROHM/ROHM.html.

Zhengyi Luo, Jinkun Cao, Rawal Khirodkar et al. · cmu

We present SimXR, a method for controlling a simulated avatar from information (headset pose and cameras) obtained from AR / VR headsets. Due to the challenging viewpoint of head-mounted cameras, the human body is often clipped out of view, making traditional image-based egocentric pose estimation challenging. On the other hand, headset poses provide valuable information about overall body motion, but lack fine-grained details about the hands and feet. To synergize headset poses with cameras, we control a humanoid to track headset movement while analyzing input images to decide body movement. When body parts are seen, the movements of hands and feet will be guided by the images; when unseen, the laws of physics guide the controller to generate plausible motion. We design an end-to-end method that does not rely on any intermediate representations and learns to directly map from images and headset poses to humanoid control signals. To train our method, we also propose a large-scale synthetic dataset created using camera configurations compatible with a commercially available VR headset (Quest 2) and show promising results on real-world captures. To demonstrate the applicability of our framework, we also test it on an AR headset with a forward-facing camera.

Julian Bedei, Lucas Koch, Kevin Badalian et al.

This work introduces a toolchain for applying Reinforcement Learning (RL), specifically the Deep Deterministic Policy Gradient (DDPG) algorithm, in safety-critical real-world environments. As an exemplary application, transient load control is demonstrated on a single-cylinder internal combustion engine testbench in Homogeneous Charge Compression Ignition (HCCI) mode, that offers high thermal efficiency and low emissions. However, HCCI poses challenges for traditional control methods due to its nonlinear, autoregressive, and stochastic nature. RL provides a viable solution, however, safety concerns, such as excessive pressure rise rates, must be addressed when applying to HCCI. A single unsuitable control input can severely damage the engine or cause misfiring and shut down. Additionally, operating limits are not known a priori and must be determined experimentally. To mitigate these risks, real-time safety monitoring based on the k-nearest neighbor algorithm is implemented, enabling safe interaction with the testbench. The feasibility of this approach is demonstrated as the RL agent learns a control policy through interaction with the testbench. A root mean square error of 0.1374 bar is achieved for the indicated mean effective pressure, comparable to neural network-based controllers from the literature. The toolchain's flexibility is further demonstrated by adapting the agent's policy to increase ethanol energy shares, promoting renewable fuel use while maintaining safety. This RL approach addresses the longstanding challenge of applying RL to safety-critical real-world environments. The developed toolchain, with its adaptability and safety mechanisms, paves the way for future applicability of RL in engine testbenches and other safety-critical settings.

Maximilian Fehrentz, Alexander Winkler, Thomas Heiliger et al.

We introduce BridgeSplat, a novel approach for deformable surgical navigation that couples intraoperative 3D reconstruction with preoperative CT data to bridge the gap between surgical video and volumetric patient data. Our method rigs 3D Gaussians to a CT mesh, enabling joint optimization of Gaussian parameters and mesh deformation through photometric supervision. By parametrizing each Gaussian relative to its parent mesh triangle, we enforce alignment between Gaussians and mesh and obtain deformations that can be propagated back to update the CT. We demonstrate BridgeSplat's effectiveness on visceral pig surgeries and synthetic data of a human liver under simulation, showing sensible deformations of the preoperative CT on monocular RGB data. Code, data, and additional resources can be found at https://maxfehrentz.github.io/ct-informed-splatting/ .

Julian Bedei, Murray McBain, Alexander Winkler et al.

Reinforcement Learning (RL) and Machine Learning Integrated Model Predictive Control (ML-MPC) are promising approaches for optimizing hydrogen-diesel dual-fuel engine control, as they can effectively control multiple-input multiple-output systems and nonlinear processes. ML-MPC is advantageous for providing safe and optimal controls, ensuring the engine operates within predefined safety limits. In contrast, RL is distinguished by its adaptability to changing conditions through its learning-based approach. However, the practical implementation of either method alone poses challenges. RL requires high variance in control inputs during early learning phases, which can pose risks to the system by potentially executing unsafe actions, leading to mechanical damage. Conversely, ML-MPC relies on an accurate system model to generate optimal control inputs and has limited adaptability to system drifts, such as injector aging, which naturally occur in engine applications. To address these limitations, this study proposes a hybrid RL and ML-MPC approach that uses an ML-MPC framework while incorporating an RL agent to dynamically adjust the ML-MPC load tracking reference in response to changes in the environment. At the same time, the ML-MPC ensures that actions stay safe throughout the RL agent's exploration. To evaluate the effectiveness of this approach, fuel pressure is deliberately varied to introduce a model-plant mismatch between the ML-MPC and the engine test bench. The result of this mismatch is a root mean square error (RMSE) in indicated mean effective pressure of 0.57 bar when running the ML-MPC. The experimental results demonstrate that RL successfully adapts to changing boundary conditions by altering the tracking reference while ML-MPC ensures safe control inputs. The quantitative improvement in load tracking by implementing RL is an RSME of 0.44 bar.

Zhengyi Luo, Jinkun Cao, Alexander Winkler et al.

We present a physics-based humanoid controller that achieves high-fidelity motion imitation and fault-tolerant behavior in the presence of noisy input (e.g. pose estimates from video or generated from language) and unexpected falls. Our controller scales up to learning ten thousand motion clips without using any external stabilizing forces and learns to naturally recover from fail-state. Given reference motion, our controller can perpetually control simulated avatars without requiring resets. At its core, we propose the progressive multiplicative control policy (PMCP), which dynamically allocates new network capacity to learn harder and harder motion sequences. PMCP allows efficient scaling for learning from large-scale motion databases and adding new tasks, such as fail-state recovery, without catastrophic forgetting. We demonstrate the effectiveness of our controller by using it to imitate noisy poses from video-based pose estimators and language-based motion generators in a live and real-time multi-person avatar use case.

Sebastian Andress, Alex Johnson, Mathias Unberath et al.

Fluoroscopic X-ray guidance is a cornerstone for percutaneous orthopaedic surgical procedures. However, two-dimensional observations of the three-dimensional anatomy suffer from the effects of projective simplification. Consequently, many X-ray images from various orientations need to be acquired for the surgeon to accurately assess the spatial relations between the patient's anatomy and the surgical tools. In this paper, we present an on-the-fly surgical support system that provides guidance using augmented reality and can be used in quasi-unprepared operating rooms. The proposed system builds upon a multi-modality marker and simultaneous localization and mapping technique to co-calibrate an optical see-through head mounted display to a C-arm fluoroscopy system. Then, annotations on the 2D X-ray images can be rendered as virtual objects in 3D providing surgical guidance. We quantitatively evaluate the components of the proposed system, and finally, design a feasibility study on a semi-anthropomorphic phantom. The accuracy of our system was comparable to the traditional image-guided technique while substantially reducing the number of acquired X-ray images as well as procedure time. Our promising results encourage further research on the interaction between virtual and real objects, that we believe will directly benefit the proposed method. Further, we would like to explore the capabilities of our on-the-fly augmented reality support system in a larger study directed towards common orthopaedic interventions.

Jonas Buchli, Farbod Farshidian, Alexander Winkler et al.

Optimal and Learning Control for Autonomous Robots has been taught in the Robotics, Systems and Controls Masters at ETH Zurich with the aim to teach optimal control and reinforcement learning for closed loop control problems from a unified point of view. The starting point is the formulation of of an optimal control problem and deriving the different types of solutions and algorithms from there. These lecture notes aim at supporting this unified view with a unified notation wherever possible, and a bit of a translation help to compare the terminology and notation in the different fields. The course assumes basic knowledge of Control Theory, Linear Algebra and Stochastic Calculus.