A low-cost robot ready for any obstacle

A low-cost robot ready for any obstacle

image: A robotic system designed by researchers at CMU and Berkeley that allows small, low-cost robots to maneuver in challenging environments.
Opinion more

Credit: Carnegie Mellon University

This little robot can go just about anywhere.

Researchers in the Carnegie Mellon University School of Computer Science and the University of California, Berkeley Design an automated system which enables a low-cost, relatively small robot to climb and descend stairs approaching its height; Traverse rocky, slippery, uneven, steep and varied terrain; walking through rock size gaps and limitations; And even work in the dark.

“Enabling small robots to climb stairs and navigate a variety of environments is critical to developing robots that will be useful in people’s homes as well as in search and rescue operations,” said Deepak Pathak, associate professor at the Robotics Institute. “This system creates a robust and adaptable robot that can perform many everyday tasks.”

The team put the robot through its paces, testing it on uneven staircases and hillsides in public parks, challenging it to walk across stepping stones and on slippery surfaces, and requiring it to climb ladders that resembled a human jumping off. obstacle. The robot quickly adapts and masters difficult terrain by relying on its vision and a small on-board computer.

The researchers trained the robot using 4,000 copies of it in a simulator, where they practiced walking and climbing in difficult terrain. The speed of the simulator allowed the robot to gain six years of experience in a single day. The simulator also stored the motor skills it learned during training in a neural network, which the researchers copied into the real robot. This approach required no manual engineering of the robot’s movements—a departure from traditional methods.

Most robotic systems use cameras to create a map of the surrounding environment and use that map to plan movements before executing them. The process is slow and can often get bogged down by inherent ambiguity, inaccuracies, or misconceptions in the mapping phase that affect planning and subsequent movements. Planning and planning is useful in systems that focus on high-level control but are not always suited to the dynamic demands of low-level skills such as walking or running over difficult terrain.

The new system bypasses the plotting and planning phases and directs vision inputs directly to the robot’s control. What the robot sees determines how it moves. Even the researchers haven’t specified how the legs should move. This technology allows the robot to react to oncoming terrain quickly and move through it effectively.

Since there is no planning or planning involved, and the movements are trained using machine learning, the robot itself can be quite low cost. The robot the team used was at least 25 times cheaper than available alternatives. The team’s algorithm has the potential to make low-cost bots more widely available.

“This system uses vision and feedback directly from the body as input to output commands for the robot’s motors,” said Ananye Agarwal, PhD, SCS. student in machine learning. “This technology allows the system to be very robust in the real world. If you slip down stairs, it can recover. It can go into unknown environments and adapt.”

This straightforward vision-to-control aspect is biologically inspired. Humans and animals use vision to move. Try running or balancing with your eyes closed. Previous research from the team showed that blind bots — robots without cameras — can navigate difficult terrain, but adding vision and relying on that vision dramatically improves the system.

The team also looked to nature for other elements in the system. For a small robot—less than a foot in this case—to climb stairs or obstacles approaching its height, it has learned to adopt the locomotion that humans use to climb higher obstacles. When a human has to raise his leg up to climb a ledge or obstacle, he uses his hips to move his leg to the side, which is called abduction and adduction, which gives him more clearance. The Pathak robotic team is doing the same, using hip abduction to tackle the hurdles that bog down some of the most advanced robotic systems on the market.

The movement of the hind legs by the four-legged animals also inspired the team. When a cat moves across obstacles, its hind legs will avoid the same items as its front legs without the benefit of a close set of eyes. Four-legged animals have a memory that enables their hind legs to track their front legs. Our system works in a similar way, said Pathak. The system’s built-in memory enables the rear legs to remember what the camera in front has seen and maneuver to avoid obstacles.

“Because there is no map and no planning, our system remembers the terrain and how the front leg moved and translates that to the back leg, quickly and without error,” said Ashish Kumar, Ph.D. student at Berkeley.

The research could be a huge step toward solving current challenges facing legged robots and bringing them into people’s homes. the paper “Move your legs around challenging terrain using selfish visionWritten by Pathak and Berkeley Professor Jitendra Malik, Agarwal and Kumar, and will be presented at the upcoming conference on robot learning in Auckland, New Zealand.

Disclaimer: AAAS and EurekAlert! Not responsible for the accuracy of the newsletters sent on EurekAlert! Through contributing organizations or for using any information through the EurekAlert system.

#lowcost #robot #ready #obstacle

Leave a Comment

Your email address will not be published. Required fields are marked *