This robotic dog can run along the beach at 9.8 feet per second

By Jim Leffman via SWNS

Jogging along the beach with your robotic dog might seem like something out of a sci-fi film, but scientists have done just that.

The robo-dog, called RaiBo, is the first to be able to navigate uneven surfaces and can bound along sand dunes at 9.8 feet (three meters) per second.

The team from the Korea Advanced Institute of Science and Technology (KAIST) used advanced neural networks to allow the dog to make judgments on the run.

It is able to adapt to various types of ground without prior information while walking at the same time.

The trained neural network controller is expected to expand the scope of application of quadrupedal walking robots by proving its robustness in changing terrain.

This includes the ability to move at high speed even on a sandy beach and walk and turn on soft grounds like an air mattress without losing balance.

Led by KAIST's Department of Mechanical Engineering, the study published in the journal Science Robotics, uses reinforcement learning.

This is an AI learning method used to create a machine that collects data on the results of various actions in an arbitrary situation. It then uses 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.

The 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 'dog.'

However the performance of the learning-based controller rapidly decreases when the actual environment has any discrepancy from the learned simulation environment.

To counter this, the team implemented 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.

By calculating the force generated from one or several contacts at each time step, the deforming terrain was efficiently simulated.

They then combined this with an artificial neural network structure that predicts ground characteristics using a recurrent neural network that analyses time-series data from the robot's sensors.

All this was incorporated into RaiBo which was built hands-on by the research team.

It was able to run 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 environment without any additional programming or revision to the controlling algorithm.

In addition, it rotated with stability at 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 developers hope it will lead to robots capable of literally thinking on their feet and performing practical tasks on a range of terrains.

First author doctoral student Soo-Young Choi said: "It has been shown that providing a learning-based controller with a close contact experience with real deforming ground is essential for the application to deforming terrain.

“The proposed controller can be used without prior information on the terrain, so it can be applied to various robot walking studies.”