Jemin+Hwangbo
-
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 248 -
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 15660