Robotic Systems Lab

Open Opportunities

RL-based Navigation with Multi-Resolution Semantic Volumetric Maps ETH Zurich Robotic Systems Lab Other organizations: Autonomous Systems Lab Wavemap is a multi-resolution volumetric mapping framework. By integrating its 3D maps with RL pipelines, we want to enable robots to navigate in complex environments. Future semantic support will enable advanced applications like urban navigation and safe traversal of hazardous terrains. Intelligent Robotics Master Thesis

GPU Accelerated Collision-Free Path-Planning for Robot Arm Control ETH Zurich Robotic Systems Lab Improve collision-free path planning for robotic arms used for waste sorting by leveraging the CUDA-accelerated parallel planning library cuRobo. Engineering and Technology Master Thesis, Semester Project

Reinforcement Learning for Excavation Planning ETH Zurich Robotic Systems Lab We aim to develop a reinforcement learning-based global excavation planner that can plan for the long term and execute a wide range of excavation geometries. The system will be deployed on our legged excavator. Intelligent Robotics Master Thesis, Semester Project

Reinforcement Learning for Particle-Based Excavation in Isaac Sim ETH Zurich Robotic Systems Lab We want to train RL agents on our new particle simulator, accelerated on the GPU via warp in Isaac sim. Intelligent Robotics Master Thesis, Semester Project

Terrain Neural Scene Reconstruction for RL ETH Zurich Robotic Systems Lab In this project, our goal is to build a practical solution for reconstructing 3D earthworks scenes using incomplete point cloud data. We plan to train an encoder-decoder neural network that can accurately recreate the missing parts of the scene. However, our main emphasis lies in creating powerful latent representations that will enable us to train reinforcement learning agents for digging tasks. Intelligent Robotics Master Thesis, Semester Project

Sample Efficienct Reinforcement Learning for Particle-Based Excavation ETH Zurich Robotic Systems Lab We want to train RL agents on our new particle simulator, accelerated on the GPU via warp in Isaac sim. Information, Computing and Communication Sciences Master Thesis, Semester Project

Multiagent RL in Terra ETH Zurich Robotic Systems Lab We want to train multiple agents in the Terra environment, a fully end-to-end GPU-accelerated environment for RL training. Information, Computing and Communication Sciences Master Thesis, Semester Project

Bridging Imitation- and Reinforcement learning to interact with Articulated Objects ETH Zurich Robotic Systems Lab Other organizations: ETH Competence Center - ETH AI Center To integrate robots into daily life, they must learn to manipulate diverse environments and objects. Recent advances in imitation learning show promise for teaching visual-motor skills, but require extensive robot-specific data. Reinforcement learning in simulation can learn robust policies in varied settings but struggles with the sim-to-real gap, especially with complex systems and camera observations. This work combines both approaches: using imitation learning to control a five-fingered hand from RGB images and reinforcement learning to control a quadruped's base and arm. Intelligent Robotics, Knowledge Representation and Machine Learning, Robotics and Mechatronics Master Thesis

Leveraging Human Motion Data from Videos for Humanoid Robot Motion Learning ETH Zurich ETH Competence Center - ETH AI Center Other organizations: Advanced Interactive Technologies, Course 6: Electrical Engineering and Computer Science, Learning and Adaptive Systems, Robotic Systems Lab The advancement in humanoid robotics has reached a stage where mimicking complex human motions with high accuracy is crucial for tasks ranging from entertainment to human-robot interaction in dynamic environments. Traditional approaches in motion learning, particularly for humanoid robots, rely heavily on motion capture (MoCap) data. However, acquiring large amounts of high-quality MoCap data is both expensive and logistically challenging. In contrast, video footage of human activities, such as sports events or dance performances, is widely available and offers an abundant source of motion data. Building on recent advancements in extracting and utilizing human motion from videos, such as the method proposed in WHAM (refer to the paper "Learning Physically Simulated Tennis Skills from Broadcast Videos"), this project aims to develop a system that extracts human motion from videos and applies it to teach a humanoid robot how to perform similar actions. The primary focus will be on extracting dynamic and expressive motions from videos, such as soccer player celebrations, and using these extracted motions as reference data for reinforcement learning (RL) and imitation learning on a humanoid robot. Engineering and Technology Master Thesis

Periodic Motion Priors for General Quadruped Locomotion Learning ETH Zurich ETH Competence Center - ETH AI Center Other organizations: Course 6: Electrical Engineering and Computer Science, Robotic Systems Lab In recent years, advancements in reinforcement learning have achieved remarkable success in quadruped locomotion tasks. Despite their similar structural designs, quadruped robots often require uniquely tailored reward functions for effective motion pattern development, limiting the transferability of learned behaviors across different models. This project proposes to bridge this gap by developing a unified, continuous latent representation of quadruped motions applicable across various robotic platforms. By mapping these motions onto a shared latent space, the project aims to create a versatile foundation that can be adapted to downstream tasks for specific robot configurations. Engineering and Technology, Information, Computing and Communication Sciences Master Thesis