Nature is the best teacher and can teach the best robot

Asian Industry Network News: According to PCMag. In nature, elegant engineering solutions abound, and experts in robotics are working to unlock their secrets, com reports.

Nature is the best teacher and can teach the best robot

It took me and the guards over 5 minutes to walk through the converted WWII-era warehouse, first through a labyrinth of dimly lit corridors and cavernous orbital modules, then through a laboratory full of spaceship skeletons, and finally Just arrived at the workbench, where the U.S. Navy is building robotic squirrels. The robotic squirrel, whose full name is the Mesoscale Robotic Motion Project (MeRLIn), can be thought of as a rodent monster. When completed this spring, the Robot will weigh about 9 to 20 pounds.

MeRLIn consists of rectangular tubes and tenth generation dog joint legs, which are then mounted on sliding aluminium struts. The dark blue 3D-printed module next to it shows its finished state: a headless, four-legged machine about the size of a Yorkshire Terrier. But when the engineers on this project kicked it off for a demo, I saw why they liked to make MeRLIn a squirrel. Although it uses a small motor and hydraulically actuated pistons, it can jump very high.

Creature Rise

MeRLIn is just one of the latest robots to find inspiration in animals. The animal kingdom is filled with many examples of clever sensing and movement, and in the world of battery-powered, power-limited autonomous robots, efficiency is key. For example, mimicking the jumping ability of kangaroos can help robots achieve the ideal balance between power consumption and performance: with each step, the hind limbs of these marsupials provide powerful stored energy, allowing kangaroos to travel long distances with less energy consumption travel.

Nature is the best teacher and can teach the best robot

Biology provides tremendous support for the most innovative robotic designs today. Check out UC Berkeley’s Salto, which took inspiration from African high jumping baby monkeys. The University of Virginia’s mantabot is modeled after the Chesapeake Bay bullnose ray. It’s easy to see why this is done. Inspiration from living things has clear advantages in design, especially when it comes to tasks that humans have a hard time adapting to and accomplishing.

From tiny flies to deep-sea fish and even microbes (some fuel cells are powered by microbial chemistry), nature is modifying and adapting living things in very efficient ways so they can adapt to a variety of environments. Millions of years of evolution have allowed animals to fly, jump, walk, and swim, sensing in an invisible spectrum. In addition, they have many more capabilities that we may not have discovered yet.

But the biorobots being built today are far from mechanical replicas of animals, and they are the direction these elegant biological solutions are headed. Now, we need to dissect these biological strategies and break down their main essence to use them to achieve our goals. Scientists and engineers build these components that move better, processors that can think deeply, and sensors that can detect with precision, and then combine them into truly useful solutions that can do difficult-to-achieve tasks in batches.

1 2 3 4 Next > page

Early challenges

If MeRLIn looks familiar, it does. In fact, MeRLIn took inspiration from other internet-famous robots, such as Boston Dynamics’ L3, Big Dog and MIT’s Cheetah, said Glen Henshaw, the project’s lead investigator. Engineers at the Naval Research Laboratory hope to develop a smaller, quieter and more agile Robot that doesn’t require two burly young Marines to carry it to inspect for potential hazards. But developing MeRLIn wasn’t a simple matter of shrinking the Robot so it could fit in a soldier’s backpack. It is also a process of understanding the function of specific gaits, why those gaits are suitable for different terrains, and how to make robots that can learn to adapt and choose the correct gait.

At MeRLIn’s workbench, control engineer Joe Hays typed a few test commands into the computer, causing the robot’s legs to twitch and jerk. After he removed the supporting aluminum pillars, MeRLIn’s single leg was able to support the brick-sized body. Control was then handed over to the hydraulic control system, and with lightning-like spasms, merRLin’s legs bounced 1 meter into the air before being redirected back onto the vertical metal track. Repeating this process 3 times, the robot hit the top of the shield after the last powerful jump, and then fell hard, breaking its legs. “Frankly, there’s a lot of animal movement that we don’t know yet,” Henshaw said. “We don’t really understand the neuromuscular system, and we’re trying to build something that doesn’t know how it should walk.”

Nature is the best teacher and can teach the best robot

The team is also working on some issues in hydraulics, but has found a good pattern of success using adaptive algorithms that detect and correct for uncertainties in hardware circuits as fast as 1 per millisecond. They believe that it is expected to jump from the ground to the table within a few months. The Minitaur, designed by Avik De and Gavin Kenneally at the University of Pennsylvania under the direction of Dan Koditschek, is the latest subcompact , Lightweight quadruped robot. The Minitaur weighs about 6 kg and can move forward with a jumping gait. But when you watch videos of it climbing stairs, jumping over walls, and jumping to unlock it, love can quickly turn to skepticism.

De and Kennelly slashed the robot’s body by using free-swinging, direct-drive legs instead of traditional gear-driven legs. The motor acts as a feedback sensor connected to the robot’s software, detecting and adjusting torque 1,000 times per second. The result is a robot that can hop slowly or quickly, climb stairs, jump up, swing its legs, and then hook a doorknob to open it. While it’s far from automated and lacks the sensors and controls that would allow it to move freely, the Minitaur’s unique, adjustable pogo stick-like action shows that flexibility is possible even without a powerful drive mechanism. It is assembled from commercially available parts.

Nature is the best teacher and can teach the best robot

“There’s clearly enough incentive to fit these people with legs, but the current state of the technology is immature and it’s very expensive,” said De, apparently referring to Boston Dynamics’ Atlas robot, which, while more capable, is expensive, The technology is complex and difficult to imitate. “We want to make robots that other people can use, so they can try it out in their own applications,” he said.

Serpentine Scheme

Howie Choset admits to being afraid of snakes. But surprisingly, almost all of his most famous works are serpentine. Josette is an associate professor at Carnegie Mellon University in the United States. He has been studying snake-like robots since he was a graduate student, and has made a series of achievements. He directs the Robotics Institute at Carnegie Mellon University, where many of the creative functions being developed are inspired by snakes. He is also editor-in-chief of the recently launched Science Robotics journal and has written a textbook on the principles of robotic motion. In his spare time, Josette also founded two companies, Hebi Robotics and Medrobotics. Among them, the latter mainly studies the advanced endoscopic surgical tool Flex Robotic System, which has been approved by the FDA in 2015.


Josette has defended whether the Flex Robotic System was inspired by snakes, saying the robot’s snake shape is based on the twists and turns of human inner space. But recent work by others has apparently been developing robots by looking at snakes and mimicking their movements, notably by collaborating with Georgia Tech physicist Dan Goldman, who studies biomechanics, through Take inspiration from crabs, turtles, cockroaches, mudskippers, and sandfish to design robots.

Josette also acknowledged that his research was influenced by one of the pioneers of bionic robotics, Robert Full, who heads the Poly-Pedal Laboratory at the University of California, Berkeley. By studying how cockroaches move and geckos climb vertical walls, Faure, Josette, and other experts have managed to boil down these secrets into general design principles that can be applied to new robotic designs. “Should we replicate biology? No, we need to ask biologists that we want the best design principles and put them to work,” Josette said.

Nature is the best teacher and can teach the best robot

Josette, Goldman, and Zoo Atlanta’s Joseph Mendelson worked together to study the rattlesnake’s movements, and ultimately characterized its violent writhing as a series of deformation waves. Applying this knowledge to the snake-like robot project, Josette’s team was able to get their robot to climb a pile of sand, a previously impossible task. Understanding how snakes change shape to suit their surroundings has also led Josette to develop new snake-like robots that can wrap around pillars and access the inside of lintels, a design he envisions could be used to explore very dangerous interior situations, such as nuclear power plants or in inaccessible archaeological sites.

“The fact is that living things are so complex, I just want to be able to apply their mechanisms to our robots,” Josette said. “But we’re not trying to replicate the capabilities and sophistication that animals have, we want to build something that’s even bigger. Mechanisms and systems of potential.” Josette’s own progress and the achievements and discoveries of his students were applied quite serendipitously to the field of development into robotics. “Evolution is blind, and there are no turning points, just a series of developments. From the outside, they look like major breakthroughs,” he said.

key intersection

Many times, engineers do not understand biological principles, which requires cooperation between engineers and biologists. At the University of Chicago, biologist Mark Westneat is studying wrasses and working with the Navy to lead them to develop WANDA, a slow-moving but very nimble underwater robot that can be used in Helps inspect hulls, docks, oil rigs, etc.

High-speed photography was an important research topic more than 20 years ago, when Westnet was just beginning his research on wrasse imaging, and the Navy has also become interested in his research since then. In a constant-current flow cell, which Westnette calls a “fish treadmill,” wrasses swim happily, using only their pectoral fins to hold a fixed position in the flow cell, while high speed The camera captures every detail of its movement at 1000 frames per second.

Combined with detailed knowledge gained by biologists dissecting wrasses, such as how fins attach to muscles, and how nerve endings in fin membranes transmit stress and tension, these photography can help scientists gain insight into how wrasses work through the Twisting the body in the water, flapping the current to propel itself. This ability allows the wrasse to stay in place and not be swept away by the current. This capability makes the wrasse an ideal model for new flexible underwater devices, said Jason Geder, lead engineer on the WANDA project. “Traditional propeller- or booster-driven underwater equipment doesn’t have this maneuverability, or turns too far,” he explained. “The wrasse is a very good fish model because if we want to be in the center of the underwater equipment, Attaching the rigid shell, we can use this pectoral fin movement to achieve similar performance.”

Nature is the best teacher and can teach the best robot

Westnet believes that new 3D photography capabilities could further enhance research. “For the fish, it’s a matter of life or death,” he said. “And for us, a better understanding of efficiency means we’ll have better power. We really want to mimic the underlying bone structure and the underlying membrane. Mechanical properties and see if we can get super high efficiency.”

The Links:   3HAB3772-75   AC2251 

Published on 10/09/2022