Researchers create an autonomously navigating robot on wheels

Fast-moving autonomous mobile robots can help deliver goods to different locations, addressing disruptions in product supply chains. Nevertheless, wheeled or legged robots alone may not be sufficient to carry out deliveries both efficiently and independently.

Researchers at ETH Zurich’s Robotic Systems Lab recently introduced a new robot design that combines the capabilities of wheeled and legged robots. This robot, introduced in a Science Robotics paper, navigates through environments using various reinforcement learning techniques, allowing it to smoothly switch between driving and walking modes and adapt to different terrains.

“The main goal of the project was to build a large-scale autonomous propulsion system for such a ground robot, with the fastest speed ever,” Joonho Lee, co-author of the paper, told Tech Xplore. “This is the result of more than five years of research in the fields of robotics, autonomous navigation and robot perception.”

The robotic system developed by Lee and his colleagues builds on an earlier robot from team CERBERUS, a team that includes researchers from the indoor drone company Flyability, which won the DARPA Subterranean Challenge in 2021. Unlike the robot developed by team CERBERUS, however, their system has a simplified design and a more advanced AI-powered navigation system.

“Traditionally, navigation planning for ground robots was done using online optimization methods,” Lee explains. “Such approaches work fine for simple wheeled or slow-moving robots, but in the case of fast-moving robots like ours (which can travel up to 20 km/h) they cannot provide navigation plans fast enough. For robots moving at a speed of 2 m/s, a delay of 0.5 seconds can result in an error of 1 m, which could lead to a catastrophic collision.”

To help their robot navigate environments autonomously, the researchers developed, trained and tested various hierarchical reinforcement learning techniques. Ultimately, they trained a neural network-based controller that can process different types of input, creating new navigation plans for the robot within milliseconds.

“Another major advantage of our approach is that our neural net controller fully understands the non-linear and complex dynamics of paw robots,” says Lee. “Because he understands how the robot behaves in different terrains and at different speeds, he can navigate the robot very efficiently.”

On smooth surfaces that are easy to move, the robot developed by ETH Zurich moves forward, using its wheels and minimizing energy consumption. In more complex terrains that are difficult or impossible to navigate using wheels, such as in the presence of steps, the robot can switch to walking mode.

The neural network-based controller, developed and trained by Lee and his colleagues, can process sensory data to determine the most efficient way for the robot to move around specific terrains. This allows the robot to effectively combine the strengths of conventional wheeled robots with those of legged robots.

“Wheeled robots are efficient, but they cannot overcome high obstacles,” Lee said. “On the other hand, leg robots are very good at overcoming obstacles and steep slopes, but their efficiency is very low because they have to actuate more than 10 joints in an irregular pattern. Normally, walking robots can only work for a maximum of 1 hour. With its mobile legs, our robot can overcome the same obstacles as normal walking robots, with an operation time of at least three times as long.”

The controller developed by Lee and his colleagues does not use classical planning and model-based control techniques. In particular, these traditional methods were found to often perform poorly in real-world conditions characterized by uncertainty and random disturbances.

Instead, the team’s controller is controlled by two artificial neural networks. These networks process data collected by sensors integrated into the robot, produce appropriate walking movements and decide in which direction the robot should move.

“To train a navigation agent, we created a special simulation environment, which resembles a computer game,” Lee said. “Our software automatically generates new ‘phases’ for the navigation controller with different complex terrains and disturbances. After several hours of training, we obtained very robust and versatile neural network controllers that can handle all kinds of rugged terrains and maze-like environments.”

A further advantage of the navigation system that controls the robot’s movements is that it is simpler than many existing controllers. One of the two neural networks it relies on focuses on planning walking movements, while the other focuses on the robot’s overall navigation. The controller also includes modules for basic terrain mapping and SLAM (simultaneous localization and mapping).

“This is the simplest navigation system design I have ever seen, while very strong neural net controllers eliminate a lot of engineering effort in system integration,” said Lee. “The actual time we spent building the navigation system itself was less than a year.”

Lee and his colleagues tested their navigation system in a series of experiments conducted in real-world environments. They found that the robot was highly reactive and performed excellently as it allowed the robot to successfully travel more than 10 km through two different European cities, namely Zurich and Seville.

In the future, the wheeled robot and navigation system introduced in this recent article could be further improved and deployed in different environments. One of their most promising applications will be the fast, reliable and autonomous delivery of goods across different terrains.

“I now want to expand this system to include multimodal inputs,” Lee added. “Currently it only relies on geometric information for navigation and walking, but in the real world there are more things we have to take into account when we walk around. For example, the robot should care about more semantic information, such as checking if the ground is wet, if it should stay on the sidewalk or grass, if there is a red traffic light on, and so on.”

© 2024 Science X Network