ANYmal for Search and Rescue Missions Robotic Systems Lab: Legged Robotics at ETH Zürich 2026-09-21 | ANYmal for Search and Rescue Missions
2022, a year of events Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-12-31 | With a wrap-up of the main events that marked 2022 for us, the RSL team wishes you a happy and eventful 2023.
ANYexo 2.0 - Experiment Video: Selective Compliance Position Control Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-12-02 | Checkout our main video of this robot prototype: youtu.be/pMKoDeaS37kDescription:The video shows an experiment comparing the reaction of a position controller to compensated and uncompensated disturbances to demonstrate the benefits of using 6-DOF interaction force sensors in combination with torque-controllable robot joint actuation. Thereby, while precisely moving and supporting a patient arm, the robot can remain compliant to collisions with bystanders or patient body parts other than the arm. Thus, increasing the safety of the device.This video is part of an article accepted for publication in IEEE Transactions on Robotics. You can find an open-access preprint under: https://www.research-collection.ethz.ch/handle/20.500.11850/584411Abstract: "We developed an exoskeleton for neurorehabilitation that covered all relevant degrees of freedom of the human arm while providing enough range of motion, speed, strength, and haptic-rendering function for therapy of severely affected (e.g., mobilization) and mildly affected patients (e.g., strength and speed). The ANYexo 2.0, uniting these capabilities, could be the vanguard for highly versatile therapeutic robotics applicable to a broad target group and an extensive range of exercises. Thus, supporting the practical adoption of these devices in clinics.The unique kinematic structure of the robot and the bio-inspired controlled shoulder coupling allowed training for most activities of daily living. We demonstrated this capability with 15 sample activities, including interaction with real objects and the own body with the robot in transparent mode. The robot’s joints can reach 200%, 398%, and 354% of the speed required during activities of daily living at the shoulder, elbow, and wrist, respectively. Further, the robot can provide isometric strength training. We present a detailed analysis of the kinematic properties and propose algorithms for intuitive control implementation."
ANYexo 2.0 - Experiment Video: Activities of Daily Living Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-12-02 | Checkout our main video of this robot prototype: youtu.be/pMKoDeaS37kDescription:This video shows performing activities of daily living while wearing the fully-actuated exoskeleton robot.This video is part of an article accepted for publication in IEEE Transactions on Robotics. You can find an open-access preprint under: https://www.research-collection.ethz.ch/handle/20.500.11850/584411Abstract: "We developed an exoskeleton for neurorehabilitation that covered all relevant degrees of freedom of the human arm while providing enough range of motion, speed, strength, and haptic-rendering function for therapy of severely affected (e.g., mobilization) and mildly affected patients (e.g., strength and speed). The ANYexo 2.0, uniting these capabilities, could be the vanguard for highly versatile therapeutic robotics applicable to a broad target group and an extensive range of exercises. Thus, supporting the practical adoption of these devices in clinics.The unique kinematic structure of the robot and the bio-inspired controlled shoulder coupling allowed training for most activities of daily living. We demonstrated this capability with 15 sample activities, including interaction with real objects and the own body with the robot in transparent mode. The robot’s joints can reach 200%, 398%, and 354% of the speed required during activities of daily living at the shoulder, elbow, and wrist, respectively. Further, the robot can provide isometric strength training. We present a detailed analysis of the kinematic properties and propose algorithms for intuitive control implementation."
ANYexo 2.0: A Fully-Actuated Upper-Limb Exoskeleton for Versatile Robot-Assisted Neurotherapy Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-12-02 | The ANYexo 2.0 is our latest prototype based on around two decades of research at the Sensory-Motor Systems Lab and Robotic Systems Lab of ETH Zürich.This video shows uncommented impressions of the main features of ANYexo 2.0 and its performance in range of motion, speed, strength, haptic transparency, and human-robot attachment system.This video is part of an article accepted for publication in IEEE Transactions on Robotics. You can find an open-access preprint under: https://www.research-collection.ethz.ch/handle/20.500.11850/584411Abstract: "We developed an exoskeleton for neurorehabilitation that covered all relevant degrees of freedom of the human arm while providing enough range of motion, speed, strength, and haptic-rendering function for therapy of severely affected (e.g., mobilization) and mildly affected patients (e.g., strength and speed). The ANYexo 2.0, uniting these capabilities, could be the vanguard for highly versatile therapeutic robotics applicable to a broad target group and an extensive range of exercises. Thus, supporting the practical adoption of these devices in clinics.The unique kinematic structure of the robot and the bio-inspired controlled shoulder coupling allowed training for most activities of daily living. We demonstrated this capability with 15 sample activities, including interaction with real objects and the own body with the robot in transparent mode. The robot’s joints can reach 200%, 398%, and 354% of the speed required during activities of daily living at the shoulder, elbow, and wrist, respectively. Further, the robot can provide isometric strength training. We present a detailed analysis of the kinematic properties and propose algorithms for intuitive control implementation."
Supporting and Stabilizing the Scapulohumeral Rhythm with a Body- or Robot-Powered Orthosis Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-11-01 | Please check out our paper on the orthosis shown in this video:Open Access: https://www.research-collection.ethz.ch/handle/20.500.11850/548152 IEEE: ieeexplore.ieee.org/document/9779351 The versatile functionality of the human upper limb is owed to the coordinated rotation of the scapula and humerus, a pattern called the scapulohumeral rhythm (SHR). Various medical conditions can alter the SHR, frequently leading to limitations in activities of daily living.However, to date, supporting the SHR in practice is often not feasible.We present a textile orthosis that assists the SHR both in stand-alone use and in combination with the ANYexo, a therapy exoskeleton, or the Myoshirt, an assistive exomuscle. The SHR Orthosis comprised a textile harness and a scapula interface that was coupled with the upper arm to promote scapular upward rotation.In a technical evaluation including four participants without impairments and one with a partial hemiparesis, the SHR Orthosis followed the desired scapular rotation with an average deviation of less than 5%, thus providing accurate support and guidance towards the physiological SHR.The SHR Orthosis substituted for up to 42.0% of the normal forces, and up to 19.6% of the tangential forces required for scapular stabilization and rotation, providing sufficient support for patients with remaining muscular function.At last, the SHR Orthosis provides practicable scapula support in daily life, during conventional therapy, and in combination with assistive and therapy robots.Music: bensound.com
Neural Scene Representation for Locomotion on Structured Terrain Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-10-03 | We propose a learning-based method to reconstruct the local terrain for locomotion with a mobile robot traversing urban environments. Using a stream of depth measurements from the onboard cameras and the robot's trajectory, the algorithm estimates the topography in the robot's vicinity. The raw measurements from these cameras are noisy and only provide partial and occluded observations that in many cases do not show the terrain the robot stands on. Therefore, we propose a 3D reconstruction model that faithfully reconstructs the scene, despite the noisy measurements and large amounts of missing data coming from the blind spots of the camera arrangement. The model consists of a 4D fully convolutional network on point clouds that learns the geometric priors to complete the scene from the context and an auto-regressive feedback to leverage spatio-temporal consistency and use evidence from the past. The network can be solely trained with synthetic data, and due to extensive augmentation, it is robust in the real world, as shown in the validation on a quadrupedal robot, ANYmal, traversing challenging settings. We run the pipeline on the robot's onboard low-power computer using an efficient sparse tensor implementation and show that the proposed method outperforms classical map representations.Published at RA-L/IROS 2022Work by David Hoeller, Nikita Rudin, Christopher Choy, Animashree Anandkumar, and Marco HutterPaper: arxiv.org/abs/2206.08077
Advanced Skills by Learning Locomotion and Local Navigation End-to-End Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-09-27 | Local navigation and locomotion of legged robots are commonly split into separate modules. In this work, we propose to combine them by training an end-to-end policy with deep reinforcement learning. Training a policy in this way opens up a larger set of possible solutions, which allows the robot to learn more complex behaviors.IROS 2022work by Nikita Rudin, David Hoeller, Marko Bjelonic, and Marco Hutterpaper: arxiv.org/abs/2209.12827project website: sites.google.com/leggedrobotics.com/end-to-end-loco-navigation/homeCredits for outdoor video: "Learning robust perceptive locomotion for quadrupedal robots in the wild" by Takahiro Miki, Joonho Lee, Jemin Hwangbo, Lorenz Wellhausen, Vladlen Koltun, Marco Hutter
Perceptive Locomotion through Nonlinear Model Predictive Control Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-08-18 | This work is currently under review, a preprint is available:researchgate.net/publication/362759412_Perceptive_Locomotion_through_Nonlinear_Model_Predictive_ControlTitle: Perceptive Locomotion through Nonlinear Model Predictive ControlAuthors:Ruben Grandia, Fabian Jenelten, Shaohui Yang, Farbod Farshidian, and Marco HutterAbstract:Dynamic locomotion in rough terrain requires accurate foot placement, collision avoidance, and planning of the underactuated dynamics of the system. Reliably optimizing for such motions and interactions in the presence of imperfect and often incomplete perceptive information is challenging. We present a complete perception, planning, and control pipeline, that can optimize motions for all degrees of freedom of the robot in real-time. To mitigate the numerical challenges posed by the terrain a sequence of convex inequality constraints is extracted as local approximations of foothold feasibility and embedded into an online model predictive controller. Steppability classification, plane segmentation, and a signed distance field are precomputed per elevation map to minimize the computational effort during the optimization. A combination of multiple-shooting, real-time iteration, and a filter-based line-search are used to solve the formulated problem reliably and at high rate. We validate the proposed method in scenarios with gaps, slopes, and stepping stones in simulation and experimentally on the ANYmal quadruped platform, resulting in state-of-the-art dynamic climbing.
Collision detection and identification for a legged manipulator Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-08-11 | To safely deploy legged robots in the real world it is necessary to provide them with the ability to reliably detect unexpected contacts and accurately estimate the corresponding contact force. In this paper, we propose a collision detection and identification pipeline for a quadrupedal manipulator. We first introduce an approach to estimate the collision time span based on band-pass filtering and show that this information is key for obtaining accurate collision force estimates. We then improve the accuracy of the identified force magnitude by compensating for model inaccuracies, unmodeled loads, and any other potential source of quasi-static disturbances acting on the robot. We validate our framework with extensive hardware experiments in various scenarios, including trotting and additional unmodeled load on the robot.Link to paper: arxiv.org/pdf/2207.14745.pdf
Keynote speech Marco Hutter ICRA 2022 Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-08-11 | ...
Locomotion Policy Guided Traversability Learning [IROS 2022] Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-08-02 | by Jonas Frey, David Hoeller, Shehryar Khattak, Marco HutterPaper Link: arxiv.org/abs/2203.15854Accepted for IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS, 2022)@inproceedings{frey2022traversability, author={Frey, Jonas and Hoeller, David and Khattak, Shehryar and Marco, Hutter}, journal={IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS, 2022)}, title={Locomotion Policy Guided Traversability Learning using Volumetric Representations of Complex Environments}, year={2022}}Video by Jonas FreyDeployed Locomotion Policy from "Learning robust perceptive locomotion for quadrupedal robots in the wild" by Takahiro Miki, Joonho Lee, Jemin Hwangbo, Lorenz Wellhausen, Vladlen Koltun, Marco HutterPaper Link: science.org/doi/10.1126/scirobotics.abk2822Youtube Video (14s-23s): youtu.be/zXbb6KQ0xV8Abstract: Despite the progress in legged robotic locomotion, autonomous navigation in unknown environments remains an open problem. Ideally, the navigation system utilizes the full potential of the robots’ locomotion capabilities while operating within safety limits under uncertainty. The robot must sense and analyze the traversability of the surrounding terrain, which depends on the hardware, locomotion control, and terrain properties. It may contain information about the risk, energy, or time consumption needed to traverse the terrain. To avoid hand-crafted traversability cost functions we propose to collect traversability information about the robot and locomotion policy by simulating the traversal over randomly generated terrains using a physics simulator. Thousand of robots are simulated in parallel controlled by the same locomotion policy used in reality to acquire 57 years of real-world locomotion experience equivalent. For deployment on the real robot, a sparse convolutional network is trained to predict the simulated traversability cost, which is tailored to the deployed locomotion policy, from an entirely geometric representation of the environment in the form of a 3D voxel-occupancy map. This representation avoids the need for commonly used elevation maps, which are error-prone in the presence of overhanging obstacles and multi-floor or low-ceiling scenarios. The effectiveness of the proposed traversability prediction network is demonstrated for path planning for the legged robot ANYmal in various indoor and natural environments.
Design and Motion Planning for a Reconfigurable Robotic Base Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-07-21 | Supplementary material to our publication: "Design and Motion Planning for a Reconfigurable Robotic Base"Preprint of the accepted paper: arxiv.org/abs/2206.15298Full text at IEEE Xplore: ieeexplore.ieee.org/document/9817632Open-Source code of the control software: github.com/leggedrobotics/swerve_steeringAbstract:A robotic platform for mobile manipulation needs to satisfy two contradicting requirements for many real-world applications: A compact base is required to navigate through cluttered indoor environments, while the support needs to be large enough to prevent tumbling or tip over, especially during fast manipulation operations with heavy payloads or forceful interaction with the environment.This paper proposes a novel robot design that fulfills both requirements through a versatile footprint. It can reconfigure its footprint to a narrow configuration when navigating through tight spaces and to a wide stance when manipulating heavy objects. Furthermore, its triangular configuration allows for high-precision tasks on uneven ground by preventing support switches.A model predictive control strategy is presented that unifies planning and control for simultaneous navigation, reconfiguration, and manipulation. It converts task-space goals into whole-body motion plans for the new robot.The proposed design has been tested extensively with a hardware prototype. The footprint reconfiguration allows to almost completely remove manipulation-induced vibrations. The control strategy proves effective in both lab experiments and during a real-world construction task.Bibtex:@ARTICLE{9817632, author={Pankert, Johannes and Valsecchi, Giorgio and Baret, Davide and Zehnder, Jon and Pietrasik, Lukasz L. and Bjelonic, Marko and Hutter, Marco}, journal={IEEE Robotics and Automation Letters}, title={Design and Motion Planning for a Reconfigurable Robotic Base}, year={2022}, volume={7}, number={4}, pages={9012-9019}, doi={10.1109/LRA.2022.3189166}}
Learning-based Localizability Estimation for Robust LiDAR Localization (supplementary) Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-07-20 | To be published at the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2022Learning-based Localizability Estimation for Robust LiDAR LocalizationJulian Nubert, Etienne Walther, Shehryar Khattak and Marco HutterPaper: arxiv.org/pdf/2203.05698.pdfCode to appear soon: github.com/leggedrobotics/L3EAbstract:LiDAR-based localization and mapping is one of the core components in many modern robotic systems due to the direct integration of range and geometry, allowing for precise motion estimation and generation of high quality maps in real-time. Yet, as a consequence of insufficient environmental constraints present in the scene, this dependence on geometry can result in localization failure, happening in self-symmetric surroundings such as tunnels. This work addresses precisely this issue by proposing a neural network-based estimation approach for detecting (non-)localizability during robot operation. Special attention is given to the localizability of scan-to-scan registration, as it is a crucial component in many LiDAR odometry estimation pipelines. In contrast to previous, mostly traditional detection approaches, the proposed method enables early detection of failure by estimating the localizability on raw sensor measurements without evaluating the underlying registration optimization. Moreover, previous approaches remain limited in their ability to generalize across environments and sensor types, as heuristic-tuning of degeneracy detection thresholds is required. The proposed approach avoids this problem by learning from a corpus of different environments, allowing the network to function over various scenarios. Furthermore, the network is trained exclusively on simulated data, avoiding arduous data collection in challenging and degenerate, often hard-to-access, environments. The presented method is tested during field experiments conducted across challenging environments and on two different sensor types without any modifications. The observed detection performance is on par with state-of-the-art methods after environment-specific threshold tuning.
Soil-Adaptive Excavation Using Reinforcement Learning Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-07-03 | RA-L/IROS 2022https://www.research-collection.ethz.ch/handle/20.500.11850/557541Abstract:In this letter, we present an excavation controller for a full-sized hydraulic excavator that can adapt online to different soil characteristics. Soil properties are hard to predict and can vary even within one scoop, which requires a controller that can adapt online to the encountered soil conditions. The objective is to fill the bucket with excavation material while respecting machine limitations to prevent stalling or lifting of the machine. To this end, we train a control policy in simulation using Reinforcement Learning (RL). The soil interactions are modeled based on the Fundamental Equation of Earth-Moving (FEE) with heavily randomized soil parameters to expose the agent to a wide range of different conditions. The agent learns to output joint velocity commands, which can be directly applied to the standard proportional valves of the real machine. We test the controller on a 12-ton excavator in different types of soils. The experiments demonstrate that the controller can adapt online to changing conditions without the explicit knowledge of the soil parameters, solely from proprioceptive observations, which are easily measurable.
TAMOLS: Terrain-Aware Motion Optimization for Legged Systems Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-06-29 | We present a model-based optimization framework that optimizes base pose and footholds simultaneously. It can generate motions in rough environments for a variety of different gaits in real time.Title: TAMOLS: Terrain-Aware Motion Optimization for Legged SystemsAuthors:Fabian Jenelten, Ruben Grandia, Farbod Farshidian, and Marco HutterIEEE: doi.org/10.1109/TRO.2022.3186804arXiv: doi.org/10.48550/arXiv.2206.14049code for mapping filters: github.com/leggedrobotics/elevation_mapping_cupyAbstract: Terrain geometry is, in general, non-smooth, non-linear, non-convex, and, if perceived through a robot-centric visual unit, appears partially occluded and noisy. This work presents the complete control pipeline capable of handling the aforementioned problems in real-time. We formulate a trajectory optimization problem that jointly optimizes over the base pose and footholds, subject to a heightmap. To avoid converging into undesirable local optima, we deploy a graduated optimization technique. We embed a compact, contact-force free stability criterion that is compatible with the non-flat ground formulation. Direct collocation is used as transcription method, resulting in a non-linear optimization problem that can be solved online in less than ten milliseconds. To increase robustness in the presence of external disturbances, we close the tracking loop with a momentum observer. Our experiments demonstrate stair climbing, walking on stepping stones, and over gaps, utilizing various dynamic gaits.Acknowledgments:This research was partially supported by the Swiss National Science Foundation (SNSF) as part of project No.188596, the European Union’s Horizon 2020 research and innovation programme under grant agreement No.780883 and No. 101016970, and the Swiss National Science Foundation through the National Centre of Competence in Research Robotics (NCCR Robotics).Voice-over by Maria Alejandra Jaimes
Offline motion libraries and online MPC for advanced mobility skills Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-06-09 | Our robot ANYmal combines offline motion libraries and online model predictive control for complex locomotion skills.Journal article published in the International Journal of Robotics Research (IJRR): journals.sagepub.com/doi/10.1177/02783649221102473Learn more about the robot at swiss-mile.comVideo by Marko Bjelonic, markobjelonic.comTitle: Offline motion libraries and online MPC for advanced mobility skillsAuthors:Marko Bjelonic, Ruben Grandia, Moritz Geilinger, Oliver Harley, Vivian S. Medeiros, VukPajovic, Edo Jelavic, Stelian Coros and Marco HutterAbstract: We describe an optimization-based framework to perform complex locomotion skills for robots with legs and wheels. The generation of complex motions over a long-time horizon often requires offline computation due to current computing constraints and is mostly accomplished through trajectory optimization (TO). In contrast, model predictive control (MPC) focuses on the online computation of trajectories, robust even in the presence of uncertainty, albeit mostly over shorter time horizons and is prone to generating nonoptimal solutions over the horizon of the task's goals. Our article's contributions overcome this trade-off by combining offline motion libraries and online MPC, uniting a complex, long-time horizon plan with reactive, short-time horizon solutions. We start from offline trajectories that can be, for example, generated by TO or sampling-based methods. Also, multiple offline trajectories can be composed out of a motion library into a single maneuver. We then use these offline trajectories as the cost for the online MPC, allowing us to smoothly blend between multiple composed motions even in the presence of discontinuous transitions. The MPC optimizes from the measured state, resulting in feedback control, which robustifies the task's execution by reacting to disturbances and looking ahead at the offline trajectory. With our contribution, motion designers can choose their favorite method to iterate over behavior designs offline without tuning robot experiments, enabling them to author new behaviors rapidly. Our experiments demonstrate complex and dynamic motions on our traditional quadrupedal robot ANYmal and its roller-walking version. Moreover, the article's findings contribute to evaluating five planning algorithms.Video content:- 00:00 Boston Dynamic's dream- 00:13 Intro- 00:20 Dance- 00:35 Summary- 01:15 Approach- 02:47 OutroAcknowledgments:This work was supported in part by the Swiss National Science Foundation (SNF) through the National Centres of Competence in Research Robotics (NCCR Robotics) and Digital Fabrication (NCCR dfab). Besides, it has been conducted as part of ANYmal Research, a community to advance legged robotics.
Team CERBERUS Wins the DARPA Subterranean Challenge Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-06-07 | This article presents the CERBERUS robotic system-of-systems, which won the DARPA Subterranean Challenge Final Event in 2021. The Subterranean Challenge was organized by DARPA with the vision to facilitate the novel technologies necessary to reliably explore diverse underground environments despite the grueling challenges they present for robotic autonomy. Due to their geometric complexity, degraded perceptual conditions combined with lack of GPS support, austere navigation conditions, and denied communications, subterranean settings render autonomous operations particularly demanding. In response to this challenge, we developed the CERBERUS system which exploits the synergy of legged and flying robots, coupled with robust control especially for overcoming perilous terrain, multi-modal and multi-robot perception for localization and mapping in conditions of sensor degradation, and resilient autonomy through unified exploration path planning and local motion planning that reflects robot-specific limitations. Based on its ability to explore diverse underground environments and its high-level command and control by a single human supervisor, CERBERUS demonstrated efficient exploration, reliable detection of objects of interest, and accurate mapping. In this article, we report results from both the preliminary runs and the final Prize Round of the DARPA Subterranean Challenge, and discuss highlights and challenges faced, alongside lessons learned for the benefit of the community.Team Website: subt-cerberus.org
Natural Intelligence: Robotic Monitoring of Habitats Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-06-06 | Within the European Union project we aim to autonomously monitor habitats.Regular monitoring of individual plant species allows for more sophisticated decision-making. The video is recorded in Perugia Italy. Special thanks for organizing the field trips:Università di Pisa - Research Center "E. Piaggio": https://www.centropiaggio.unipi.it/More about the EU-Project Website: https://www.nih2020.eu/homeMore about the Robotic Systems Lab:Website: https://rsl.ethz.ch/LinkedIn: linkedin.com/company/leggedroboticsYouTube: youtube.com/user/leggedrobotics/videosMusic by Bensound.com
Rocky VI: Robot Training Montage Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-05-24 | Finally, after the first Rocky movie in 1976, the Robotic Systems Lab presents a continuation of the iconic series. Our transformer robot visited Philly in 2022 as part of the International Conference on Robotics and Automation.Credits: Marko Bjelonic, Hendrik Kolvenbach, Takahiro Miki, Nikita Rudin, Vassilios Tsounis, Maria Vittoria Minniti, Julian Nubert, Lorenz Wellhausen, Jonhoo Lee, Marco HutterLearn more at swiss-mile.com/.
Deep Measurement Updates for Bayes Filters - ICRA 2022 presentation Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-05-19 | Presentation for the IEEE International Conference on Robotics and Automation (ICRA) 2022Johannes Pankert, Maria Vittoria Minniti, Lorenz Wellhausen, Marco HutterPaper: arxiv.org/abs/2112.00380 Measurement update rules for Bayes filters often contain hand-crafted heuristics to compute observation probabilities for high-dimensional sensor data, like images. In this work, we propose the novel approach Deep Measurement Update (DMU) as a general update rule for a wide range of systems. DMU has a conditional encoder-decoder neural network structure to process depth images as raw inputs. Even though the network is trained only on synthetic data, the model shows good performance at evaluation time on real-world data. With our proposed training scheme primed data training , we demonstrate how the DMU models can be trained efficiently to be sensitive to condition variables without having to rely on a stochastic information bottleneck. We validate the proposed methods in multiple scenarios of increasing complexity, beginning with the pose estimation of a single object to the joint estimation of the pose and the internal state of an articulated system. Moreover, we provide a benchmark against Articulated Signed Distance Functions(A-SDF) on the RBO dataset as a baseline comparison for articulation state estimation.
Graph-based Multi-sensor Fusion for Consistent Localization of Autonomous Construction Robots (Talk) Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-05-14 | Presentation for the IEEE International Conference on Robotics and Automation (ICRA) 2022Julian Nubert, Shehryar Khattak and Marco HutterPaper: arxiv.org/pdf/2203.01389.pdfCode: github.com/leggedrobotics/GMFCLAbstract:Enabling autonomous operation of large-scale construction machines, such as excavators, can bring key benefits for human safety and operational opportunities for applications in dangerous and hazardous environments. To facilitate robot autonomy, robust and accurate state-estimation remains a core component to enable these machines for operation in a diverse set of complex environments. In this work, a method for multimodal sensor fusion for robot state-estimation and localization is presented, enabling operation of construction robots in realworld scenarios. The proposed approach presents a graph-based prediction-update loop that combines the benefits of filtering and smoothing in order to provide consistent state estimates at high update rate, while maintaining accurate global localization for large-scale earth-moving excavators. Furthermore, the proposed approach enables a flexible integration of asynchronous sensor measurements and provides consistent pose estimates even during phases of sensor dropout. For this purpose, a dualgraph design for switching between two distinct optimization problems is proposed, directly addressing temporary failure and the subsequent return of global position estimates. The proposed approach is implemented on-board two Menzi Muck walking excavators and validated during real-world tests conducted in representative operational environments.
Robotic Sechseläuten 2022 Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-04-29 | The Sechseläuten ("The six o'clock ringing of the bells") is a traditional spring holiday in the Swiss city of Zürich and the parade had this year a robotic guest of honor!Press images: drive.google.com/drive/folders/1ZSvrAsGkJHalgpxhyMBvXZpMw0x8ZXzG?usp=sharingCredits: Marko Bjelonic, Marco Tranzatto, Carl Ziegler, Maria Trodella, Markus Montenegro, Pius Kolb, Alessandro Fulciniti, Marco HutterLearn more at swiss-mile.com/.
Transformer robot featured at Nvidias GTC 2022 Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-04-13 | The video shows snippets of our wheeled-legged robot featured at Nvidia's GTC 2022 keynote.by David Hoeller, Nikita Rudin, Marko Bjelonic, Victor Klemm, Eric Vollenweider, Joonho Lee, Marco HutterFull robot video: youtu.be/kEdr0ARq48ACredit by Nvidia: youtu.be/39ubNuxnrK8Learn more about the robot at swiss-mile.com
Advanced Skills through Multiple Adversarial Motion Priors in Reinforcement Learning Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-03-22 | Nvidia presented parts of this work at GTC 2022, revealing our humanoid-quadruped transformer!Title: Advanced Skills through Multiple Adversarial Motion Priors in Reinforcement LearningAuthors: Eric Vollenweider, Marko Bjelonic, Victor Klemm, Nikita Rudin, Joonho Lee and Marco HutterPaper submitted to IEEE/RSJ International Conference on Intelligent Robots and Systems in Kyoto.Preprint: arxiv.org/abs/2203.14912Abstract: In recent years, reinforcement learning (RL) has shown outstanding performance for locomotion control of highly articulated robotic systems. Such approaches typically involve tedious reward function tuning to achieve the desired motion style. Imitation learning approaches such as adversarial motion priors aim to reduce this problem by encouraging a pre-defined motion style. In this work, we present an approach to augment the concept of adversarial motion prior-based RL to allow for multiple, discretely switchable styles. We show that multiple styles and skills can be learned simultaneously without notable performance differences, even in combination with motion data-free skills. Our approach is validated in several real-world experiments with a wheeled-legged quadruped robot showing skills learned from existing RL controllers and trajectory optimization, such as ducking and walking, and novel skills such as switching between a quadrupedal and humanoid configuration. For the latter skill, the robot is required to stand up, navigate on two wheels, and sit down. Instead of tuning the sit-down motion, we verify that a reverse playback of the stand-up movement helps the robot discover feasible sit-down behaviors and avoids tedious reward function tuning.Note: The following parts of the video are sped up:- Door opening and closing when the robot stands inside the elevator between 00:20 and 00:21 (+200%)- In-between the standing up and sitting down sequence between 00:49 and 01:10 (+200% only the navigation on two legs)- Reaction with the crowd between 1:41 and 1:50 (+150%)- Last drone footage after 2:01 (+200%) Learn more at swiss-mile.com
Reconstructing Occluded Elevation Information in Terrain Maps With Self-Supervised Learning Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-03-11 | This paper presents a self-supervised learning based method to reconstruct the occluded elevation map.Published in: IEEE Robotics and Automation Letters ( Volume: 7, Issue: 2, April 2022)IEEE Xplore: ieeexplore.ieee.org/abstract/document/9676411arXiv: arxiv.org/abs/2109.07150 Code: github.com/mstoelzle/solving-occlusion Abstract:Accurate and complete terrain maps enhance the awareness of autonomous robots and enable safe and optimal path planning. Rocks and topography often create occlusions and lead to missing elevation information in the Digital Elevation Map (DEM). Currently, these occluded areas are either fully avoided during motion planning or the missing values in the elevation map are filled-in using traditional interpolation, diffusion or patch-matching techniques. These methods cannot leverage the high-level terrain characteristics and the geometric constraints of line of sight we humans use intuitively to predict occluded areas. We introduce a self-supervised learning approach capable of training on real-world data without a need for ground-truth information to reconstruct the occluded areas in the DEMs. We accomplish this by adding artificial occlusion to the incomplete elevation maps constructed on a real robot by performing ray casting. We first evaluate a supervised learning approach on synthetic data for which we have the full ground-truth available and subsequently move to several real-world datasets. These real-world datasets were recorded during exploration of both structured and unstructured terrain with a legged robot, and additionally in a planetary scenario on Lunar analogue terrain. We state a significant improvement compared to the baseline methods both on synthetic terrain and for the real-world datasets. Our neural network is able to run in real-time on both CPU and GPU with suitable sampling rates for autonomous ground robots. We motivate the applicability of reconstructing occlusion in elevation maps with preliminary motion planning experiments.
Marsupial Walking-and-Flying Robotic Deployment for Collaborative Exploration Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-03-10 | Marsupial Walking-and-Flying Robotic Deployment for Collaborative Exploration of Unknown EnvironmentsPaolo De Petris, Mihir Dharmadhikari, Huan Nguyen, Nikhil Khedekar, Kostas Alexis are with the Autonomous Robots Lab, NTNU, Norway. autonomousrobotslab.comShehryar Khattak, Gabriel Waibel, Markus Montenegro, Marco Hutter are with the Robotic Systems Lab, ETH Zurich, Switzerland. https://rsl.ethz.ch/Abstract:This work contributes a marsupial robotic system-of-systems involving a legged and an aerial robot capableof collaborative mapping and exploration path planning thatexploits the heterogeneous properties of the two systems and theability to selectively deploy the aerial system from the groundrobot. Exploiting the dexterous locomotion capabilities and longendurance of quadruped robots, the marsupial combination canexplore within large-scale and confined environments involvingrough terrain. However, as certain types of terrain or verticalgeometries can render any ground system unable to continue itsexploration, the marsupial system can –when needed– deploy theflying robot which, by exploiting its 3D navigation capabilities,can undertake a focused exploration task within its endurancelimitations. Focusing on autonomy, the two systems can co-localize and map together by sharing LiDAR-based maps andplan exploration paths individually, while a tailored graph searchonboard the legged robot allows it to identify where and whenthe ferried aerial platform should be deployed. The system isverified within multiple experimental studies demonstrating theexpanded exploration capabilities of the marsupial system-of-systems and facilitating the exploration of otherwise individuallyunreachable areas.
Graph-based Multi-sensor Fusion for Consistent Localization of Autonomous Construction Robots (Sup.) Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-03-08 | Published at the IEEE International Conference on Robotics and Automation (ICRA) 2022Graph-based Multi-sensor Fusion for Consistent Localization of Autonomous Construction RobotsJulian Nubert, Shehryar Khattak and Marco HutterPaper: arxiv.org/pdf/2203.01389.pdfCode: github.com/leggedrobotics/GMFCLAbstract:Enabling autonomous operation of large-scale construction machines, such as excavators, can bring key benefits for human safety and operational opportunities for applications in dangerous and hazardous environments. To facilitate robot autonomy, robust and accurate state-estimation remains a core component to enable these machines for operation in a diverse set of complex environments. In this work, a method for multimodal sensor fusion for robot state-estimation and localization is presented, enabling operation of construction robots in realworld scenarios. The proposed approach presents a graph-based prediction-update loop that combines the benefits of filtering and smoothing in order to provide consistent state estimates at high update rate, while maintaining accurate global localization for large-scale earth-moving excavators. Furthermore, the proposed approach enables a flexible integration of asynchronous sensor measurements and provides consistent pose estimates even during phases of sensor dropout. For this purpose, a dualgraph design for switching between two distinct optimization problems is proposed, directly addressing temporary failure and the subsequent return of global position estimates. The proposed approach is implemented on-board two Menzi Muck walking excavators and validated during real-world tests conducted in representative operational environments.
Whole-Body MPC and Dynamic Occlusion Avoidance: A Maximum Likelihood Visibility Approach Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-03-04 | This paper introduces a novel approach for whole-body motion planning and dynamic occlusion avoidance. The proposed approach reformulates the visibility constraint as a likelihood maximization of visibility probability. In this formulation, we augment the primary cost function of a whole-body model predictive control scheme through a relaxed log barrier function yielding a relaxed log-likelihood maximization formulation of visibility probability. The visibility probability is computed through a probabilistic shadow field that quantifies point light source occlusions. We provide the necessary algorithms to obtain such a field for both 2D and 3D cases. We demonstrate 2D implementations of this field in simulation and 3D implementations through real-time hardware experiments. We show that due to the linear complexity of our shadow field algorithm to the map size, we can achieve high update rates, which facilitates onboard execution on mobile platforms with limited computational power. Lastly, we evaluate the performance of the proposed MPC reformulation in simulation for a quadrupedal mobile manipulator.Paper pre-print available at: arxiv.org/pdf/2203.02221.pdf
A Collision-Free MPC for Whole-Body Dynamic Locomotion and Manipulation Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-03-01 | Abstract: In this paper, we present a real-time whole-body planner for collision-free legged mobile manipulation. We enforce both self-collision and environment-collision avoidance as soft constraints within a Model Predictive Control (MPC) scheme that solves a multi-contact optimal control problem. By penalizing the signed distances among a set of representative primitive collision bodies, the robot is able to safely execute a variety of dynamic maneuvers while preventing any self-collisions. Moreover, collision-free navigation and manipulation in both static and dynamic environments are made viable through efficient queries of distances and their gradients via a euclidean signed distance field. We demonstrate through a comparative study that our approach only slightly increases the computational complexity of the MPC planning. Finally, we validate the effectiveness of our framework through a set of hardware experiments involving dynamic mobile manipulation tasks with potential collisions, such as locomotion balancing with the swinging arm, weight throwing, and autonomous door opening.In IEEE International Conference on Robotics and Automation (ICRA) 2022 in Philadelphia (PA), USAAuthors: Jia-Ruei Chiu, Jean-Pierre Sleiman, Mayank Mittal, Farbod Farshidian, Marco HutterThis research was supported in part by the Swiss National Science Foundation through the National Centre of Competence in Research Robotics (NCCR Robotics), and in part by TenneT.Paper pre-print available at: arxiv.org/pdf/2202.12385.pdf
Deep Measurement Updates for Bayes Filters Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-02-17 | This video is supplementary material to the RA-L 01/2022 publication "Deep Measurement Updates for Bayes Filters" by Johannes Pankert, Maria Vittoria Minniti, Lorenz Wellhausen, and Marco Hutter arXiv: arxiv.org/abs/2112.00380IEEE RA-L: ieeexplore.ieee.org/document/9618810
A Reconfigurable Leg for Walking Robots Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-02-15 | We present the design of a robotic leg that can seamlessly switch between a spring-suspended-, and unsuspended configuration. Switching is realized by a lightweight mechanism that exploits the alternative configuration of the two-link leg. This allows us to embed additional functionality in the leg to increase the performance in relation to the locomotion task.Title: A Reconfigurable Leg for Walking RobotsAuthors: Fang Nan, Hendrik Kolvenbach and Marco Hutter IEEE Robotics and Automation Letters (RA-L) Volume: 7, Issue: 2, April 2022ieeexplore.ieee.org/document/9667211DOI: 10.1109/LRA.2021.3139379Also available here:https://www.research-collection.ethz.ch/handle/20.500.11850/522681This work was conducted as part of ANYmal Research, a community to advance legged robotics. This research was supported by the Swiss National Science Foundation through the National Centre of Competence in Research Robotics (NCCR Robotics).For more information visit:rsl.ethz.ch
Combining Learning-Based Locomotion with Model-Based Manipulation for Legged Mobile Manipulators Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-02-01 | Deep reinforcement learning produces robust locomotion policies for legged robots over challenging terrains. To date, few studies have leveraged model-based methods to combine these locomotion skills with the precise control of manipulators. Here, we incorporate external dynamics plans into learning-based locomotion policies for mobile manipulation. We train the base policy by applying a random wrench sequence on the robot base in simulation and add the noisified wrench sequence prediction to the policy observations. The policy then learns to counteract the partially-known future disturbance. The random wrench sequences are replaced with the wrench prediction generated with the dynamics plans from model predictive control to enable deployment. We show zero-shot adaptation for manipulators unseen during training. On the hardware, we demonstrate stable locomotion of legged robots with the prediction of the external wrench.Link to the publication: ieeexplore.ieee.org/abstract/document/9684679
Learning robust perceptive locomotion for quadrupedal robots in the wild Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-01-19 | We present a perceptive locomotion controller for quadrupedal robots that combines fast locomotion and exceptional robustness on challenging terrain.Learn more about this work on our project website: leggedrobotics.github.io/rl-perceptivelocoOur version of paper: leggedrobotics.github.io/rl-perceptiveloco/assets/pdf/wild_anymal.pdfAuthors: Takahiro Miki, Joonho Lee, Jemin Hwangbo, Lorenz Wellhausen, Vladlen Koltun, Marco HutterPublished in Science Robotics: science.org/doi/10.1126/scirobotics.abk2822Video by Takahiro Miki0:00 Introduction0:37 Hiking experiment1:31 DARPA SubT Challenge1:57 Monkey research2:08 How it works2:59 How to train3:33 What belief state captures4:32 Unreliable maps5:09 Stairs5:31 Summary
Adaptive CLF-MPC with application to quadrupedal robots Robotic Systems Lab: Legged Robotics at ETH Zürich 2022-01-06 | Modern robotic systems are endowed with superior mobility and mechanical skills that make them suited to be employed in real-world scenarios, where interactions with heavy objects and precise manipulation capabilities are required. For instance, legged robots with high payload capacity can be used in disaster scenarios to remove dangerous material or carry injured people. It is thus essential to develop planning algorithms that can enable complex robots to perform motion and manipulation tasks accurately. In addition, online adaptation mechanisms with respect to new, unknown environments are needed. In this work, we impose that the optimal state-input trajectories generated by Model Predictive Control (MPC) satisfy the Lyapunov function criterion derived in adaptive control for robotic systems. As a result, we combine the stability guarantees provided by Control Lyapunov Functions (CLFs) and the optimality offered by MPC in a unified adaptive framework, yielding an improved performance during the robot's interaction with unknown objects.We validate the proposed approach in simulation and hardware tests on a quadrupedal robot carrying un-modeled payloads and pulling heavy boxes.Link to the publication: ieeexplore.ieee.org/document/9618829
Quadrupedal Robots for Planetary Exploration Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-12-15 | Dr. Hendrik Kolvenbach presents the research results performed at the Robotic Systems Lab of ETH Zurich on dynamically walking quadrupedal robots for planetary exploration. He focuses on the locomotion aspects and experimental validation. This lecture was held as part of the Lecture Series on Space Research and Exploration at ETH Zurich (07.12.2021). 07:39 Motivation13:01 Dynamically walking quadrupeds21:23 Low-gravity locomotion30:49 Locomotion on dry, granular media37:38 Using limbs to probe the environment41:49 Inspecting subterranean environments47:10 Objectives in lunar exploration53:17 ESA Resource Challenge1:00:45 Q&A Further details can be found in the following document: doi.org/10.3929/ethz-b-000489008
Team CERBERUS: DARPA Subterranean Challenge Technical Approach and Lessons Learned Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-12-07 | Team CERBERUS' DARPA Subterranean Challenge Technical Approach and Lessons Learned. We outline our team's approach with respect to the robotic systems legged and flying mobility concepts, methods for resilient multi-modal and multi-robot localization and mapping, autonomy and especially exploration path planning, artifact detection and localization on the map, as well as communications and networking. Finally, we outline our competitive runs and results especially during the Prize Run of the DARPA Subterranean Challenge finals. Presenters:* Kostas Alexis* Marco Tranzatto* Shehryar Khattak* Mihir Dharmadhikari* Mihir Kulkarni* Samuel ZimmermannTeam CERBERUS info page: http://www.subt-cerberus.orgNote: There is a typo error in 26min: We mention 40,007 labels total because we also accounted for labels for the cellphone artifact. Visual detection was however not reliable thus the associated row ("Cellphone - 3556 labels") is not shown in the table.
The Future of Robotic Mobility Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-12-03 | It’s a car, quadruped, and humanoid. Our wheeled-legged robot performs exceptional locomotion skills with reinforcement learning driving at speeds of up to 6.2 m/s (22.32 km/h or 13.87 miles/h), overcoming obstacles, and standing up on two legs!With both legs and wheels, our robot outperforms state-of-the-art wheeled delivery platforms as well as lightweight delivery drones. It is the only solution capable of carrying tools, materials, goods, and sensors over long distances with energy efficiency and speed while overcoming challenging obstacles like steps and stairs and enabling seamless navigation in indoor and outdoor urban environments.Learn more: swiss-mile.comVideo by Marko Bjelonic (markobjelonic.com)Video content:- 00:00 A car- 00:20 A quadruped- 00:26 A humanoid- 00:35 Real-life transformer- 01:11 Outro
Recovery Controller for a Quadrupedal Robot using Deep Reinforcement Learning Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-11-18 | Master's thesis by Joonho Lee (scholar.google.com/citations?user=6Htb7swAAAAJ&hl=en)Paper link: arxiv.org/abs/1901.07517
Towards Robotic Harvesting: Mapping, Localization, Planning and Control for a Legged Tree Harvester Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-11-08 | Full title: Towards Autonomous Robotic Precision Harvesting: Mapping, Localization, Planning and Control for a Legged Tree HarvesterWe present an integrated system for performing precision harvesting missions using a legged harvester (HEAP) in a confined, GPS denied forest environment. The mission starts with a human mapping the area of interest using a custom-made sensor module. Subsequently, a human expert selects the trees for harvesting. The sensor module is then mounted on the machine and used for localization within the given map. Upon reaching the approach pose, the machine grabs a tree with a general-purpose gripper. This process repeats for all the trees selected by the operator. Our system has been tested on a testing field with tree trunks and in a natural forest.The paper can be found on arxiv: arxiv.org/abs/2104.10110Code available at:Planning & path tracking: github.com/leggedrobotics/se2_navigationHeightmap generation: github.com/ANYbotics/grid_map/tree/master/grid_map_pclLocalization: github.com/leggedrobotics/icp_localizationTree detection: github.com/leggedrobotics/tree_detectionThis research was partially supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme grant agreement No 852044 and through the SNSF National Centre of Competence in Digital Fabrication (NCCR dfab).
ANYmal meets Monkeys in the Wild of South Africa Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-10-14 | This project investigates the interaction between robots and animals, in particular, the quadruped ANYmal and wild vervet monkeys. We will test whether robots can be tolerated but also socially accepted in a group of vervets. We will evaluate whether social bonds are created between them and whether vervets trust knowledge from robots.Inkawu Vervet Project and University of Lausanne: Erica van de Waal, Charlotte Canteloup (inkawuvervetproject.weebly.com, https://www.unil.ch/dee/en/home/menuinst/research--education/research/research-groups/van-de-waal-group.html)Robotic researchers: Joonho Lee, Samuel Zimmermann, Markus Montenegro, Marco Hutter (https://rsl.ethz.ch/)Drone footage by Lukas Schad.Video editing by Marko Bjelonic (markobjelonic.com/).
Learning to Walk in Minutes Using Massively Parallel Deep RL Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-10-05 | We present a training set-up that achieves fast policy generation for real-world robotic tasks by using massive parallelism on a single workstation GPU. The parallel approach allows training policies for flat terrain in under 4 minutes, and in 20 minutes for uneven terrain. Paper accepted to CoRL 2021.Code and paper: leggedrobotics.github.io/legged_gym
ANYmal at DARPA SubT final run Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-09-27 | This is a collection of all video snippets of the final run of the DARPA subT challenge broadcasted by DARPA featuring the ANYmal robots of team CERBERUS.
Autonomous subterranean exploration with the ANYmal C Robot inside the Hagerbach underground mine Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-09-10 | autonomousrobotslab.com/exploration-planning.html
Perceptive Model Predictive Control for Continuous Mobile Manipulation Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-07-01 | This video is supplementary material to the work "Perceptive Model Predictive Control for Continuous Mobile Manipulation" published in RA-L 2020.IEEE paper link:ieeexplore.ieee.org/abstract/document/9145591Open Access Version:https://www.research-collection.ethz.ch/handle/20.500.11850/439902Paper Abstract:A mobile robot needs to be aware of its environment to interact with it safely. We propose a recedinghorizon control scheme for mobile manipulators that tracks task space reference trajectories. It uses visual information to avoid obstacles and haptic sensing to control interaction forces. Additional constraints for mechanical stability and joint limits are met. The proposed method is faster than state of the art sampling based planners, open-source available and can be implemented on a broad class of robots. We validate the method both in simulation and through extensive hardware experiments with a multitude of mobile manipulation platforms. The resulting software package is released with this paper.Force Torque Sensor used: botasys.com/rokubi
Grasping and Object Reorientation for Autonomous Construction of Stone Structures Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-06-25 | In this paper, we present a method for grasp planning and object manipulation that enables the world’s first autonomous assembly of a large-scale stone wall with an unmanned hydraulic excavator system. Our method utilizes point clouds and mesh surface reconstruction of stones in order toplan grasp configurations with a 2-jaw gripper mounted on the excavator. Besides considering the geometry of the stone to sample force closure grasps, the grasp planner also takes into account collision constraints during object pick and place, informed by a LiDAR based point cloud map. Furthermore, we show an approach to reorient arbitrarily shaped objects that are not feasible to be directly placed at the desired location without violating collision constraints. Using a physics engine, we find a settled intermediate pose that allows direct placement and is reachable from the initial stone pose. The applicability of the proposed grasp planning method is demonstrated with the construction of a dry stone wall composed of over one hundred boulders using an autonomous excavator. We show a high primary grasp success rate (82.2 %) and illustrate how the system recovers from slippage by relocating the object and re-planning the grasp correspondingly.Authors: Martin Wermelinger, Ryan Johns, Fabio Gramazio, Matthias Kohler, and Marco HutterIEEE Robotics and Automation LettersFull paper: ieeexplore.ieee.org/document/9392354
Cat-like Jumping and Landing of Legged Robots in Low-gravity Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-06-16 | We show that deep reinforcement learning can be used to learn policies for legged locomotion control tasks encountered in space exploration. such as three-dimensional re-orientation and landing of a quadruped robot exploring low-gravity celestial bodies. Using sim-to-real transfer, we deploy trained policies in the real world on the SpaceBok robot placed on an experimental testbed designed for two-dimensional micro-gravity experiments.Title: Cat-like Jumping and Landing of Legged Robots in Low-gravity Using Deep Reinforcement LearningAuthors: Nikita Rudin, Hendrik Kolvenbach, Vassilios Tsounis and Marco Hutter IEEE Transactions on Robotics (Early Access):ieeexplore.ieee.org/document/9453856DOI: 10.1109/TRO.2021.3084374Also available here:https://www.research-collection.ethz.ch/handle/20.500.11850/490128arxiv.org/abs/2106.09357This work was supported by the European Space Agency (ESA) and Airbus DS in the framework of the Network Partnering Initiative 481-2016.For more information visit:rsl.ethz.ch
Traversing Steep and Granular Martian Analog Slopes With a Dynamic Quadrupedal Robot Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-06-10 | We present experimental work on traversing steep, granular slopes with the dynamically walking quadrupedal robot SpaceBok. We validate static and dynamic locomotion with two different foot types (point foot and passive-adaptive planar foot) on Mars analog slopes of up to 25°(the maximum of the testbed). Tests were performed at the ExoMars locomotion validation testbed located at RUAG, Switzerland. The soil used in this experiment is called "ES-3", a martian soil analog (coarse, poorly-sorted, well-graded sand). Authors: Hendrik Kolvenbach, Philip Arm, Elias Hampp, Alexander Dietsche, Valentin Bickel, Benjamin Sun, Christoph Meyer and Marco HutterField Robotics, 2021arxiv.org/abs/2106.01974Part of the thesis "Quadrupedal Robots for Planetary Exploration":https://www.research-collection.ethz.ch/handle/20.500.11850/489008For more information visit:rsl.ethz.chruag.comThis work would not have been possible without the generous offer to perform testing at RUAG Space and the support from Philipp Oettershagen and his team. This work has been supported by the European Space Agency (ESA) and Airbus DS in the framework of the Network Partnering Initiative 481-2016 and the Swiss Space Center as part of the Call for Ideas 2019.
Meet Dyana - Dynamic Animatronic Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-06-07 | Focus Roll-out 2021 - Project DyanaPresented in German, English subtitles are available.Website: https://dyana.ethz.ch/LinkedIn: linkedin.com/company/dyana-ethInstagram: instagram.com/dyana_ethTHE PROJECTWe are DyanaOur vision is to build a dynamic animatronic robot that bridges the gap between animatronics and mobile quadrupeds.By combining the world of realistic but fixed animatronics and mobile but technical robotics we aim to build a robot that brings our cat-like character to life.
Model Predictive Robot-Environment Interaction Control for Mobile Manipulation Tasks Robotic Systems Lab: Legged Robotics at ETH Zürich 2021-06-01 | Modern, torque-controlled service robots can reg-ulate contact forces when interacting with their environment.Model Predictive Control (MPC) is a powerful method to solvethe underlying control problem, allowing to plan for whole-body motions while including different constraints imposed bythe robot dynamics or its environment. However, an accuratemodel of the robot-environment is needed to achieve a satisfyingclosed-loop performance. Currently, this necessity underminesthe performance and generality of MPC in manipulation tasks.In this work, we combine an MPC-based whole-body controllerwith two adaptive schemes, derived from online system identi-fication and adaptive control. As a result, we enable a generalmobile manipulator to interact with unknown environments,without any need for re-tuning parameters or pre-modeling theinteracting objects. In combination with the MPC controller,the two adaptive approaches are validated and benchmarkedwith a ball-balancing manipulator in a door opening and objectlifting task.Paper: arxiv.org/abs/2106.04202
