This Inexpensive Four-Legged Robot Can Take On Almost Any Obstacle

Could a compact low-cost robot that can take on obstacles change how we view exploration? Researchers at the Carnegie Mellon University's School of Computer Science and the University of California, Berkeley have come up with a four-legged robot that can climb and get down the stairs nearly its own height.

This four-legged robot can also take on rocky, uneven, slippery, and steep terrain. It is also able to ably walk across gaps and function in the dark.

Deepak Pathak, a professor from Robotics Institute said that "empowering small robots to climb stairs and handle a variety of environments is crucial to developing robots that will be useful in people's homes as well as search-and-rescue operations." This adaptable kind of robot could perform "many everyday tasks."

Carnegie Mellon University

Testing the four-legged beast

The robot was tested on uneven stairs and in public parks, where it was challenged to walk over stepping stones and slippery surface. To function with maximum utility, this robot relies on vision and a fitted tiny computer.

4,000 clones of the robot were trained in a simulator, where they practiced walking and climbing on difficult terrain. Owing to the simulator's speed, the robot was able to gain six years worth of experience in just a day.

Also read: Robots Struggle To Hold Things, So Scientists Built A Creepy Tentacle Gripper

Training received in mastering motor skills in a neural network was copied on to the real robot, allowing it to became adept quickly.

Carnegie Mellon University

Most robotic systems depend on cameras to make a map of surrounding environment before making any moves. This process is not only slow, it can also sometimes be inaccurate. That kind of intense mapping and planning is not required for skills like walking and taking over challenging terrains.

Also read: In Conversation With Ameca, The World's Most Advanced Humanoid Robot

This four-legged robot decides how to move based on what it sees; there is no specification on how the legs ought to be moved from the researchers' end.

Carnegie Mellon University

Owing to this, the robot can move around much faster and efficiently. The robot's dependence on machine learning instead of mapping makes it a low-cost option - about 25 times cheaper than available alternatives, the team claimed.

What do you think about this four-legged robot? Let us know in the comments below. For more in the world of technology and science, keep reading Indiatimes.com.

References

A Low-Cost Robot Ready for Any Obstacle. (n.d.). Carnegie Mellon School of Computer Science. https://www.cs.cmu.edu/news/2022/visual-locomotion

Aupperlee, A. (2022, November 16). A low-cost robot ready for any obstacle. https://techxplore.com/news/2022-11-low-cost-robot-ready-obstacle.html