%52%61%69%42%6f
-
RAIBO Runs over Walls with Feline Agility... Ready for Effortless Search over Mountaineous and Rough Terrains
< Photo 1. Research Team Photo (Professor Jemin Hwangbo, second from right in the front row) >
KAIST's quadrupedal robot, RAIBO, can now move at high speed across discontinuous and complex terrains such as stairs, gaps, walls, and debris. It has demonstrated its ability to run on vertical walls, leap over 1.3-meter-wide gaps, sprint at approximately 14.4 km/h over stepping stones, and move quickly and nimbly on terrain combining 30° slopes, stairs, and stepping stones. RAIBO is expected to be deployed soon for practical missions such as disaster site exploration and mountain searches.
Professor Jemin Hwangbo's research team in the Department of Mechanical Engineering at our university announced on June 3rd that they have developed a quadrupedal robot navigation framework capable of high-speed locomotion at 14.4 km/h (4m/s) even on discontinuous and complex terrains such as walls, stairs, and stepping stones.
The research team developed a quadrupedal navigation system that enables the robot to reach its target destination quickly and safely in complex and discontinuous terrain.
To achieve this, they approached the problem by breaking it down into two stages: first, developing a planner for planning foothold positions, and second, developing a tracker to accurately follow the planned foothold positions.
First, the planner module quickly searches for physically feasible foothold positions using a sampling-based optimization method with neural network-based heuristics and verifies the optimal path through simulation rollouts.
While existing methods considered various factors such as contact timing and robot posture in addition to foothold positions, this research significantly reduced computational complexity by setting only foothold positions as the search space. Furthermore, inspired by the walking method of cats, the introduction of a structure where the hind feet step on the same spots as the front feet further significantly reduced computational complexity.
< Figure 1. High-speed navigation across various discontinuous terrains >
Second, the tracker module is trained to accurately step on planned positions, and tracking training is conducted through a generative model that competes in environments of appropriate difficulty.
The tracker is trained through reinforcement learning to accurately step on planned plots, and during this process, a generative model called the 'map generator' provides the target distribution.
This generative model is trained simultaneously and adversarially with the tracker to allow the tracker to progressively adapt to more challenging difficulties. Subsequently, a sampling-based planner was designed to generate feasible foothold plans that can reflect the characteristics and performance of the trained tracker.
This hierarchical structure showed superior performance in both planning speed and stability compared to existing techniques, and experiments proved its high-speed locomotion capabilities across various obstacles and discontinuous terrains, as well as its general applicability to unseen terrains.
Professor Jemin Hwangbo stated, "We approached the problem of high-speed navigation in discontinuous terrain, which previously required a significantly large amount of computation, from the simple perspective of how to select the footprint positions. Inspired by the placements of cat's paw, allowing the hind feet to step where the front feet stepped drastically reduced computation. We expect this to significantly expand the range of discontinuous terrain that walking robots can overcome and enable them to traverse it at high speeds, contributing to the robot's ability to perform practical missions such as disaster site exploration and mountain searches."
This research achievement was published in the May 2025 issue of the international journal Science Robotics.
Paper Title: High-speed control and navigation for quadrupedal robots on complex and discrete terrain, (https://www.science.org/doi/10.1126/scirobotics.ads6192)YouTube Link: https://youtu.be/EZbM594T3c4?si=kfxLF2XnVUvYVIyk
2025.06.04 View 8123 -
KAIST’s RAIBO2 becomes the World’s First Robo-dog to Successfully Complete a Full-course Marathon
KAIST's quadrupedal walking robot "RAIBO", which can run seamlessly on sandy beaches, has now evolved into "RAIBO2"and achieved the groundbreaking milestone by becomeing the world's first quadrupedal robot to successfully complete a full-course marathon in an official event.
< Photo 1. A group photo of RAIBO2 and the team after completing the full-course marathon >
KAIST (President Kwang Hyung Lee) announced on the 17th of November that Professor Je Min Hwangbo's research team of the Department of Mechanical Engineering participated in the 22nd Sangju Dried-Persimmon Marathon and completed the full-course race (42.195 km) with a time of 4 hours 19 minutes and 52 seconds.
< Photo 2. RAIBO2 after completing the full-course marathon with its official record presented on the photo wall >
The Sangju Dried Persimmon Marathon is known for its challenging course featuring two 50 m elevation climbs, each at the 14 km and 28 km marks, making it defficult for amateur runners. This made it an especially demanding challenge for the walking robot, as unexpected losses in efficiency could occur.
< Photo 3. RAIBO2 with the completion medal around its neck >
To prepare RAIBO2, Professor Hwangbo's team developed a walking controller using reinforcement learning algorithms within their proprietary simulation environment "RaiSim". This simulator allowed the team to simlate diverse terrains such as slopes, stairs, and icy roads to ensure stable walking performance.
In particular, RAIBO2's high torque transparency joint mechanism enable the robot to efficiently harvest energy on the downhill slopes to regain some of the energy used in climbing up steep hill.
In addition, the stability of the robot was greatly improved through the collaboration with RAION ROBOTICS Inc., a company founded by the researchers from Professor Hwangbo’s lab.
< Figure 1. Conceptual diagram of power flow employed by the quadrupedal robot >
< Figure 2. The process of leg posture change of RAIBO2 walking at the most efficient walking speed of 3 m/s. By reducing the ground contact speed of the feet, the collision energy loss was reduced, and by minimizing the slipperiness of the foot upon contact, the body's kinetic energy was maintained towards the direction of the movement. >
Due to the nature of walking, pedal robots must employ highly complex systems that can withstand periodic vibrations from the frequent impacts that occur upon contact with the ground surface. Immediately after development, high efficiency was already recorded in short-distance experiments in the laboratory at the beginning of the year, but the manufacturing technology of RAION ROBOTICS significantly bolstered RAIBO's performance in running safely for a prolonged time of more than 4 hours among random pack of people in an actual marathon.
Compared to previous studies on improving walking efficiency, where external parts or software could not be changed and only limited improvements were made in some areas, Professor Hwangbo’s research team cited the fact that they were able to comprehensively solve problems by developing all steps and parts in-house, including mechanism design, electrical design, software, and artificial intelligence, as a key factor in improving efficiency.
Following the development of RAIBO1, the research team developed RAIBO2 and optimized all aspects of the robot. In particular, the team integrated the motor driver circuitry directly into the robot to minimize actuator losses and increase the control bandwidth, greatly improving walking efficiency and stability.
< Photo 4. RAIBO2 running the full-course marathon along human participants >
Choongin Lee, a Ph.D. Student that co-first author of the studies on RAIBO, said, “Through the marathon project, we demonstrated that RAIBO2 has the walking performance to stably execute services such as delivery and patrol in urban environments with many people and random objects,” and “In follow-up research, we will add autonomous navigation functions to RAIBO and strive to achieve the world’s best walking performance in mountainous and disaster environments.”
< Photo 5. RAIBO2 and co-first authors of the related research at the Ph.D. program of the Department of Mechanical Engineering at KAIST. (From left) Choongin Lee, Donghoon Youm, and Jeongsoo Park >
This research was conducted with the support of Samsung Electronics Future Technology Promotion Center and RAION ROBOTICS Inc.
2024.11.17 View 13151 -
KAIST’s Beach-Roaming Quadrupedal Robot “RAIBO” to Run a Marathon!
“RAIBO”, KAIST’s four-legged robot featuring remarkable agility even on challenging terrains like sandy beaches, is now set to be the first in the world to complete a full marathon.
< Photo 1. A group photo of the research team of Professor Je Min Hwangbo (second from the right in the front row) of the Department of Mechanical Engineering who participated in the marathon event at 2024 Geumsan Insam Festival last September >
On the 17th of November, KAIST (represented by President Kwang Hyung Lee) announced that Professor Je Min Hwangbo’s team from the Department of Mechanical Engineering has developed an upgraded version, “RAIBO2,” which will take on the full 42.195-kilometer course at the "Sangju Dried-Persimmon Marathon".
This is over double the previous maximum distance achieved by quadruped robots, which was limited to around 20 kilometers. The KAIST team has successfully developed a robot that can walk continuously for 43 kilometers on a single charge, completing the course in 4 hours and 40 minutes by following a GPS-guided path on the university’s main athletic field. Through this marathon, the team aims to demonstrate RAIBO2’s walking performance in an actual urban environment.
Previously, most measurements of walking robots’ travel distances were confined to controlled laboratory conditions or theoretical data. This marathon challenge is thus significant in that the robot will run alongside the general public in a real urban setting, marking the first attempt to validate the practical potential of four-legged robots in real environments.
Quadruped robots have shown advantages in challenging terrains, such as ice, sand, and mountainous areas, where they can walk stably. However, limited travel distance and running time have long been obstacles to wider applications.
< Figure 1. Conceptual diagram of power flow employed by the quadruped robot >
Professor Hwangbo’s team designed every component of the robot, from its actuators to its mechanical structure, to overcome these limitations. Notably, they implemented an efficient walking control system based on reinforcement learning using their proprietary dynamic simulator “RaiSim”.
The team also collected and analyzed walking data from outdoor environments, creating a model to address walking losses. This model was then used to iteratively improve walking efficiency over one full year.
< Figure 2. The leg posture change process of RAIBO2 walking at the most efficient walking speed of 3 m/s. By reducing the ground contact speed of the feet, the collision energy loss was reduced, and by minimizing the slipperiness of the foot upon contact, the body's kinetic energy was maintained towards the direction of the movement. >
This is the team’s second attempt. Their first was during the marathon event at “Geumsan Insam Festival” in September when the robot’s battery ran out at the 37-kilometer mark, falling short of completion. The battery drained 10 kilometers earlier than expected due to frequent speed changes as the robot adjusted to the pacing of other runners on the course.
Following the initial attempt, the team focused on technical improvements for a successful finish. They enhanced control efficiency by implementing joint stiffness control directly onto the motor actuator and increased battery capacity by 33% by refining the internal structure. These improvements enabled the robot to cover a maximum distance of 67 kilometers on straight paths.
< Figure 3. Data from completing 43 km on a single charge at the main sports field on campus. Left) GPS data of the driving course used during autonomous running. Middle) Covered distance by the hour. Right) Energy consumption by module >
Choongin Lee, a Ph.D. student of RAI Lab who is one of the co-first author of this study, explained, “Our comprehensive analysis of walking losses in terms of mechanics, electrical systems, and walking methods was crucial to improving walking efficiency. This research marks an important milestone in extending the operating range of quadruped robots to urban environments.”
< Photo 2. A Photo from Practice Run>
This research was supported by the Samsung Electronics Future Technology Development Center and RAION ROBOTICS Co., Ltd.
< Photo 3. A Photo from Practice Run >
2024.11.15 View 9385 -
KAIST’s Robo-Dog “RaiBo” runs through the sandy beach
KAIST (President Kwang Hyung Lee) announced on the 25th that a research team led by Professor Jemin Hwangbo of the Department of Mechanical Engineering developed a quadrupedal robot control technology that can walk robustly with agility even in deformable terrain such as sandy beach.
< Photo. RAI Lab Team with Professor Hwangbo in the middle of the back row. >
Professor Hwangbo's research team developed a technology to model the force received by a walking robot on the ground made of granular materials such as sand and simulate it via a quadrupedal robot. Also, the team worked on an artificial neural network structure which is suitable in making real-time decisions needed in adapting to various types of ground without prior information while walking at the same time and applied it on to reinforcement learning. The trained neural network controller is expected to expand the scope of application of quadrupedal walking robots by proving its robustness in changing terrain, such as the ability to move in high-speed even on a sandy beach and walk and turn on soft grounds like an air mattress without losing balance.
This research, with Ph.D. Student Soo-Young Choi of KAIST Department of Mechanical Engineering as the first author, was published in January in the “Science Robotics”. (Paper title: Learning quadrupedal locomotion on deformable terrain).
Reinforcement learning is an AI learning method used to create a machine that collects data on the results of various actions in an arbitrary situation and utilizes that set of data to perform a task. Because the amount of data required for reinforcement learning is so vast, a method of collecting data through simulations that approximates physical phenomena in the real environment is widely used.
In particular, learning-based controllers in the field of walking robots have been applied to real environments after learning through data collected in simulations to successfully perform walking controls in various terrains.
However, since the performance of the learning-based controller rapidly decreases when the actual environment has any discrepancy from the learned simulation environment, it is important to implement an environment similar to the real one in the data collection stage. Therefore, in order to create a learning-based controller that can maintain balance in a deforming terrain, the simulator must provide a similar contact experience.
The research team defined a contact model that predicted the force generated upon contact from the motion dynamics of a walking body based on a ground reaction force model that considered the additional mass effect of granular media defined in previous studies.
Furthermore, by calculating the force generated from one or several contacts at each time step, the deforming terrain was efficiently simulated.
The research team also introduced an artificial neural network structure that implicitly predicts ground characteristics by using a recurrent neural network that analyzes time-series data from the robot's sensors.
The learned controller was mounted on the robot 'RaiBo', which was built hands-on by the research team to show high-speed walking of up to 3.03 m/s on a sandy beach where the robot's feet were completely submerged in the sand. Even when applied to harder grounds, such as grassy fields, and a running track, it was able to run stably by adapting to the characteristics of the ground without any additional programming or revision to the controlling algorithm.
In addition, it rotated with stability at 1.54 rad/s (approximately 90° per second) on an air mattress and demonstrated its quick adaptability even in the situation in which the terrain suddenly turned soft.
The research team demonstrated the importance of providing a suitable contact experience during the learning process by comparison with a controller that assumed the ground to be rigid, and proved that the proposed recurrent neural network modifies the controller's walking method according to the ground properties.
The simulation and learning methodology developed by the research team is expected to contribute to robots performing practical tasks as it expands the range of terrains that various walking robots can operate on.
The first author, Suyoung Choi, said, “It has been shown that providing a learning-based controller with a close contact experience with real deforming ground is essential for application to deforming terrain.” He went on to add that “The proposed controller can be used without prior information on the terrain, so it can be applied to various robot walking studies.”
This research was carried out with the support of the Samsung Research Funding & Incubation Center of Samsung Electronics.
< Figure 1. Adaptability of the proposed controller to various ground environments. The controller learned from a wide range of randomized granular media simulations showed adaptability to various natural and artificial terrains, and demonstrated high-speed walking ability and energy efficiency. >
< Figure 2. Contact model definition for simulation of granular substrates. The research team used a model that considered the additional mass effect for the vertical force and a Coulomb friction model for the horizontal direction while approximating the contact with the granular medium as occurring at a point. Furthermore, a model that simulates the ground resistance that can occur on the side of the foot was introduced and used for simulation. >
2023.01.26 View 22177