Anthropomorphic bipedal robots are by far the most impressive, but far from the most practical. Four-legged robots are much more stable. Check out the top of the most unusual, powerful, weirdest and most amazing four-legged robots so far! Here we go! The strongest four-legged robot in the world is HyQReal, created by developers at the Italian  Institute of Technology. The hydraulically powered robot weighs 130 kg, is 1.3 m long and just under a meter high. A 48-volt battery powers two computers, one for visual data processing and one for motion calculation. Four electric motors drive hydraulic pumps, which create pressure in the system. 

The power of hydraulic drives is enough to tow a Piaggio P180 Avanti passenger plane weighing 3.3 tons. In tests, the robot was able to pull the airplane 10 meters. The robust, reliable and energy-efficient robot is fully autonomous. HyQReal's aluminum frame is lined with Kevlar,   fiberglass, and plastic. The robot is designed to support people in emergency situations. The most practical four-legged robot is probably the Vision 60 from Ghost Robotics. This is the first robot-dog to get a real job on a U.S. military base. 


Although the price of the robot was not disclosed, we know that the military chose Vision 60, which they named Q-UGV, or "four-legged unmanned vehicle," precisely for its low cost, reliability, and practicality. The robot is not afraid of dirt, can overcome steep stony slopes, crawl on a half-crouch in   the grass, run rather briskly on flat terrain and overcome water obstacles. In this case, the design of the robot is modular, and any of its components that have failed, can be replaced by a  spare in just a few minutes. Q-UGV is designed to patrol the perimeter of military bases, the robot can be controlled using a virtual reality headset. If the operator sees intruders near the base, he can order them to leave the protected area using a radio attached to the robot. 



The first alien four-legged robot could be Boston Dynamics' Spot robot. We talked in detail about   the company's robots in the issue at the link in the hint. Here we look at the prospects of a robot for Mars exploration. The idea to use robot dogs came from an underground orienteering competition won by an upgraded version of the Au-Spot robot. The robot dog got networked sensors and software that allowed it to safely and autonomously scan, navigate and map the environment in underground tunnels. Au-Spot processes input from LIDAR, visual, thermal and motion sensors to create three-dimensional maps.

 The future Mars Dog also uses artificial intelligence to learn what structures to avoid and identify objects that might be of scientific interest, while a communications module allows the robot to transmit data to the surface while it explores underground. Spot's direct competitor, the ANYmal C (Enimal-C) robot, was the world's first   robopod to demonstrate no water phobia. The quadropod is able to move at a speed  of one meter per second and perform other tasks for more than two hours. The robot weighs 50 kg and is 80 cm high and can carry a load of up to 10 kg. 



To navigate, the robot uses a bunch of LIDAR at the top of the body and depth cameras mounted on each side. This allows ANYmal C to produce a three-dimensional and highly accurate map of  the environment in real time. A computer with two six-core Intel i7 processors and another dual-core  Intel i7 processor is responsible for sensor information processing and other calculations.   In addition, there is an interface on the top that can be used to connect an NVIDIA Jetson  AGX Xavier board to the robot. 

This interface can also be used to extend the robot's capabilities. The Cheetah 3 robot was the first and so far the only completely blind robot able to move   on difficult terrain, including stairs, jump and maintain equilibrium under external influences. MIT engineers deprived the robot of cameras, lidars and radars, but equipped it with two   algorithms that help the robot move by touch. The first tells Cheetah 3 about the condition  of each limb 20 times per second, using data from gyroscopes and accelerometers. It calculates the chances of each foot making contact with the ground, the probability of force generated when   the foot steps on the ground, and the probability that the foot will not touch the ground where expected.

Also Read...

The Future Of Solid State Wind Energy

 Based on this information, the robot calculates how and when to move its feet. A second model prediction control algorithm calculates the foot and body position options half a second   into the future. This helps the robot maintain its balance even with a sudden external impact. At the same time, the robot's knees bend both ways, so it doesn't need to turn around. By the way, the Cheetah 3's little brothers, the Mini Cheetah robots, are probably some of the most nimble quadropods in the world. The 9-kilogram research robots can run at almost 9 km/h, do somersaults, and land carefully when falling from heights. If you want to learn more about these and other robots, don't forget to subscribe to our channel. But the most nimble among the quadrupeds is probably the robot Dyret.



 It may still be  concept, but it is the world's first quadropod capable of adapting to the terrain as much as   possible. In each new habitat, the robot changes the length of its legs to move in the most optimal way, and remembers the results, accumulating life experience for the future. The robot can use the   built-in camera and sensors in its legs to see uneven terrain and sense how tough the surface is. The technology also allows the robot to learn to recover if one of its legs is damaged or broken. What's also distinctive about Dyret is that its ability to transform doesn't require complex construction. 

In the future, engineers want to develop a simple and reliable fully transformable robot for unpredictable environment, which will be able to change not only legs, but also its body. And a group of Google researchers has created a robot that can learn to walk on its own with minimal human intervention. Deep learning with reinforcement helped Google's robot Rainbow Dash learn to walk on different surfaces by trial and error. 

A special feature of the study was that the robot learned to walk not in a simulation, but immediately in person. It took Rainbow Dash 1.5 hours to learn to walk on a flat floor from scratch. It took 5.5   and 4.5 hours to learn to walk on a soft mattress and a relief mat, respectively. Fortunately, if we combine the skills of all the above robots, we will find a robot that is independent, self-taught, capable of self-training, performing tasks even for the blind and adapting to achieve goals. But back to the amazing quadrupeds. 

Swiss engineers have taught a robot to skate deftly on ice for the first time. Thanks to software developed to quickly create and train robots of various designs, the robot Skaterbot was able to learn wave-like movements on its own, allowing it not only to skate forward, but also to turn in the right direction. The world's most unusual four-legged robot by design is ALPHRED2. Engineers at the University of California, Berkeley, have created a robot whose legs can bend in both directions and also  rotate relative to each other. It can, for example, use three legs to move around and one to push a door, or two legs to stand still and two more to move small objects. 


Each leg of the robot has three segments and a small folding section at the end. There are two such sections on two legs and one on two more. They can synchronously rotate and allow the robot to stand on two legs, so it does not fall on its side. The cheapest four-legged robot in the world is Solo 8, developed by a team from New York University's Tandon School of Engineering.

 The open-source   robot for schools and institutes is similar in functionality to its more expensive counterparts, particularly thanks to its torque-controlled motors. The robot can bounce, get up after a fall, and flip over. All parts of the robot can be printed on a 3D printer or bought in a store.The robot can be used for research in the field of movement on different types of terrain, training with reinforcement of complex and dynamic behaviors, and others. 

Unfortunately, nothing has been heard for a long time about the Centauro robot, a promising development for rescue operations. The robot was designed to combine the best of   both worlds - the stability of a quadropod and the agility of a humanoid. The robot's legs  could change height and move apart for greater stability. 

Whereas strong arms were designed to free the way and even work with human tools. The 90 kg, one and a half meter long robot was controlled by an operator through a system that mimicked the movements of human body parts without using an exoskeleton. The system uses an ASUS Xtion PRO depth camera for motion detection, with data sent to the Open Pose algorithm. This allows you to control the robot without any training. Subscribe to the channel, like the video and do not miss new episodes about robots and technologies of the future!