JLJoonho Lee

Joonho Lee

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Recognitions

Top Cited Young Scholar (2026)
Top Cited Scholar (2026)

Joonho Lee has a total of 135 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Bridging vision-language reasoning and learned control for legged navigation in the wild

    by Takahiro Miki, Robin Schmid, Manthan Patel, Turcan Tuna, Julia Richter, Julian Nubert, Joonho Lee, Matthias Müller, Marco Hutter

    Autonomous navigation in diverse real-world environments remains one of the greatest challenges in robotics. It demands both robust low-level motion control to handle diverse terrain and high-level reasoning to interpret environmental cues. We present ROAM (Robust Open-world Autonomous Mobility), a system that enables legged robots to perform kilometer-scale navigation in complex alpine and urban settings. ROAM combines the world knowledge of a vision-language model (VLM), pretrained on internet-scale data, with robust locomotion and local planning policies learned in simulation via reinforcement learning (RL). The VLM provides common-sense reasoning and semantic interpretation of the scene, assessing risks, interpreting signs, and producing semantic labels with associated navigation costs. An open-vocabulary segmentation module converts these labels into a dense, real-time risk map, which conditions two pretrained RL policies responsible for agile locomotion and local navigation: effectively coupling high-level VLM reasoning with low-level physical control. Real-world experiments showed the system’s ability to autonomously navigate a wide range of complex real-world environments. The robot completed kilometer-long hikes with minimal intervention, interpreted signs to avoid danger, and evaluated its environment to identify safe areas on which to step. This work represents a step toward the reliable deployment of systems in unstructured environments. However, it also underscores the challenges that remain in VLM reliability and spatial reasoning, indicating that foundational models are a crucial component but not yet a complete solution for navigation in the wild.

  • Gait Optimization for Underwater Legged Robots Using Data-Driven Hydrodynamic Modeling and Reinforcement Learning

    by Seokyong Song, Taesik Kim, Seungmin Kim, Joonho Lee, Son-Cheol Yu

    Precise close-contact inspections are critical in underwater environments, where complex dynamics and biofouling present significant challenges for conventional vehicles. To address these issues, this study proposes a Reinforcement Learning (RL)-based framework to optimize the gait of an underwater legged robot for accurate and stable locomotion. A simulation environment was developed by modeling buoyancy, added mass, seabed interactions, and data-driven leg hydrodynamic forces. The control framework was designed for close-contact inspections by using a structured action space and stability-promoting rewards. The trained policy was validated through both simulation and real-world experiments, demonstrating effective mitigation of hydrodynamic disturbances. Results showed reduced pitch and altitude oscillations during high-speed forward walking and improved heading accuracy in curved trajectories.


  • Learning Fast, Tool-Aware Collision Avoidance for Collaborative Robots

    by Joonho Lee, Yunho Kim, Seokjoon Kim, Quan Nguyen, Youngjin Heo

    Ensuring safe and efficient operation of collaborative robots in human environments is challenging, especially in dynamic settings where both obstacle motion and tasks change over time. Current robot controllers typically assume full visibility and fixed tools, which can lead to collisions or overly conservative behavior. In our work, we introduce a toolaware collision avoidance system that adjusts in real time to different tool sizes and modes of tool-environment interaction. Using a learned perception model, our system filters out robot and tool components from the point cloud, reasons about occluded area, and predicts collision under partial observability. We then use a control policy trained via constrained reinforcement learning to produce smooth avoidance maneuvers in under 10 milliseconds. In simulated and real-world tests, our approach outperforms traditional approaches (APF, MPPI) in dynamic environments, while maintaining sub-millimeter accuracy. Moreover, our system operates with approximately 60 % lower computational cost compared to a state-of-the-art GPU-based planner. Our approach provides modular, efficient, and effective collision avoidance for robots operating in dynamic environments. We integrate our method into a collaborative robot application and demonstrate its practical use for safe and responsive operation.


  • Open Access

    Door-to-Door Parcel Delivery From Supply Point to User’s Home With Heterogeneous Robot Team: The euROBIN First-Year Robotics Hackathon

    by Alejandro Suarez, Rainer Kartmann, Daniel Leidner, Luca Rossini, Johann Huber, Carlos Azevedo, Quentin Rouxel, Marko Bjelonic, Antonio Gonzalez-Morgado, Christian Dreher, et al.

    Logistics and service operations involving parcel preparation, delivery, and unpacking from a supply point to a user’s home could be carried out completely by robots in the near future, taking advantage of the capabilities of the different robot morphologies for the logistics, outdoor, and domestic environments. The use of robots for parcel delivery can contribute to the goals of sustainability and reduced emissions by exploiting their different locomotion modalities (wheeled, legged, and aerial). This article reports the development and results obtained from the first robotics hackathon celebrated as part of the European Robotics and Artificial Intelligence Network involving eight robotic platforms in three domains: 1) an industrial robotic arm for parcel preparation at the supply point, 2) a Centauro robot, a dual-arm aerial manipulator, and a wheeled-legged quadruped for parcel transportation, and 3) two humanoid robots and two commercial mobile manipulators for parcel delivery and unpacking in domestic scenarios. The article describes the joint operation and the evaluation scenario, the features and capabilities of the robots, particularly those involved in the realization of the tasks, and the lessons learned.


  • Scalable Multi-Robot Cooperation for Multi-Goal Tasks Using Reinforcement Learning

    by Tianxu An, Joonho Lee, Marko Bjelonic, Flavio De Vincenti, Marco Hutter

    Coordinated navigation of an arbitrary number of robots to an arbitrary number of goals is a big challenge in robotics, often hindered by scalability limitations of existing strategies. This letter introduces a decentralized multi-agent control system using neural network policies trained in simulation. By leveraging permutation invariant neural network architectures and model-free reinforcement learning, our policy enables robots to prioritize varying numbers of collaborating robots and goals in a zero-shot manner without being biased by ordering or limited by a fixed capacity. We validate the task performance and scalability of our policies through experiments in both simulation and real-world settings. Our approach achieves a 10.3% higher success rate in collaborative navigation tasks compared to a policy without a permutation invariant encoder. Additionally, it finds near-optimal solutions for multi-robot navigation problems while being two orders of magnitude faster than an optimization-based centralized controller. We deploy our multi-goal navigation policies on two wheeled-legged quadrupedal robots, which successfully complete a series of multi-goal navigation missions.


  • Open Access

    Wing-strain-based flight control of flapping-wing drones through reinforcement learning

    by Taewi Kim, Insic Hong, Sunghoon Im, Seungeun Rho, Minho Kim, Yeonwook Roh, Changhwan Kim, Jieun Park, Daseul Lim, Doohoe Lee, et al.


  • Open Access

    Fly-by-Feel: Wing Strain-based Flight Control of Flapping-Wing Drones through Reinforcement Learning

    by Daeshik Kang, Seungyong Han, Je-Sung Koh, Taewi Kim, Insic Hong, Sunghoon Im, Seungeun Rho, Minho Kim, Yeonwook Roh, Changhwan Kim, et al.

    Although drone technology has progressed significantly, replicating the dynamic control and wind-sensing abilities of biological flights is still beyond our reach. Biological studies have revealed that insect wings are equipped with mechanoreceptors known as campaniform sensilla, which detect complex aerodynamic loads critical for flight agility. By leveraging robotic experiments designed to mimic these biological systems, we confirmed that wing strain provides crucial information about the drone's attitude, as well as the direction and velocity of the wind. We introduce a novel wing strain-based flight controller, termed 'fly-by-feel'. This methodology employs the aerodynamic forces exerted on a flapping drone's wings to deduce vital flight data, such as attitude and airflow without accelerometers and gyroscopic sensors. Our empirical approach spanned five key experiments: initially validating the wing strain sensor system for state information provision, followed by a single degree of freedom (1 DOF) control in changing winds, a two degrees of freedom (2 DOF) control for gravitational attitude adjustment, a test for position control in windy conditions, and finally, demonstrating precise flight path manipulation in a windless condition using only wing strain sensors. We have successfully demonstrated control of a flapping drone in a various environment using only wing strain sensors, with the aid of reinforcement learning-driven flight controller. The fly-by-feel system holds the potential to revolutionize autonomous drone operations, providing enhanced adaptability to environmental shifts. This will be beneficial across varied applications, from gust resistance to wind-assisted flight, paving the way toward the next generation of resilient and autonomous flying robots.


  • Top Cited Papers

    Learning robust autonomous navigation and locomotion for wheeled-legged robots

    by Joonho Lee, Marko Bjelonic, Alexander Reske, Lorenz Wellhausen, Takahiro Miki, Marco Hutter

    Autonomous wheeled-legged robots have the potential to transform logistics systems, improving operational efficiency and adaptability in urban environments. Navigating urban environments, however, poses unique challenges for robots, necessitating innovative solutions for locomotion and navigation. These challenges include the need for adaptive locomotion across varied terrains and the ability to navigate efficiently around complex dynamic obstacles. This work introduces a fully integrated system comprising adaptive locomotion control, mobility-aware local navigation planning, and large-scale path planning within the city. Using model-free reinforcement learning (RL) techniques and privileged learning, we developed a versatile locomotion controller. This controller achieves efficient and robust locomotion over various rough terrains, facilitated by smooth transitions between walking and driving modes. It is tightly integrated with a learned navigation controller through a hierarchical RL framework, enabling effective navigation through challenging terrain and various obstacles at high speed. Our controllers are integrated into a large-scale urban navigation system and validated by autonomous, kilometer-scale navigation missions conducted in Zurich, Switzerland, and Seville, Spain. These missions demonstrate the system’s robustness and adaptability, underscoring the importance of integrated control systems in achieving seamless navigation in complex environments. Our findings support the feasibility of wheeled-legged robots and hierarchical RL for autonomous navigation, with implications for last-mile delivery and beyond.


  • Seeing Through the Grass: Semantic Pointcloud Filter for Support Surface Learning

    by Anqiao Li, Chenyu Yang, Jonas Frey, Joonho Lee, Cesar Cadena, Marco Hutter

    Mobile ground robots require perceiving and understanding their surrounding support surface to move around autonomously and safely. The support surface is commonly estimated based on exteroceptive depth measurements, e.g., from LiDARs. However, the measured depth fails to align with the true support surface in the presence of high grass or other penetrable vegetation. In this work, we present the semantic pointcloud flter (SPF), a convolutional neural network (CNN) that learns to adjust LiDAR measurements to align with the underlying support surface. The SPF is trained in a semi-self-supervised manner and takes as an input a LiDAR pointcloud and RGB image. The network predicts a binary segmentation mask that identifes the specifc points requiring adjustment, along with estimating their corresponding depth values. To train the segmentation task, 464 distinct images are manually labeled into rigid and non-rigid terrain. The depth estimation task is trained in a self-supervised manner by utilizing the future footholds of the robot to estimate the support surface based on a Gaussian process. Our method can correctly adjust the support surface prior to interacting with the terrain and is extensively tested on the quadruped robot ANYmal. We show the qualitative benefts of SPF in natural environments for elevation mapping and traversability estimation compared to using raw sensor measurements and existing smoothing methods. Quantitative analysis is performed in various natural environments, and an improvement by 48% RMSE is achieved within a meadow terrain.


  • Open Access

    Evaluation of Constrained Reinforcement Learning Algorithms for Legged Locomotion

    by Joonho Lee, Lukas Schroth, Victor Klemm, Marko Bjelonic, Alexander Reske, Marco Hutter

    Shifting from traditional control strategies to Deep Reinforcement Learning (RL) for legged robots poses inherent challenges, especially when addressing real-world physical constraints during training. While high-fidelity simulations provide significant benefits, they often bypass these essential physical limitations. In this paper, we experiment with the Constrained Markov Decision Process (CMDP) framework instead of the conventional unconstrained RL for robotic applications. We perform a comparative study of different constrained policy optimization algorithms to identify suitable methods for practical implementation. Our robot experiments demonstrate the critical role of incorporating physical constraints, yielding successful sim-to-real transfers, and reducing operational errors on physical systems. The CMDP formulation streamlines the training process by separately handling constraints from rewards. Our findings underscore the potential of constrained RL for the effective development and deployment of learned controllers in robotics.