ANYmal Solves Robot Dog Walking Problems

It can even make it through deep snow. Also, Google's dragonscale roof and DARPA's new night vision glasses.

Google's Dragonscale Solar Roof

Google is erecting two new buildings in Mountain View, California, and they have a unique, if not scaley, design. Instead of your standard shingle, the roof has a "dragonscale" solar skin. It consists of 90,000, 3-foot by 5-foot silver solar panels that can generate about seven megawatts. The roof is part of the company's effort to operate carbon-free by 2030.

Google partnered with Swiss company SunStyle, which created the highly textured prismatic glass shingles. According to Google, prismatic glass can "trap" light that eludes traditional flat solar panels. They are also covered in an anti-glare coating to make them less reflective, which means less glare problems for drivers and pilots. 

The roof's curvature will optimize the amount of time the dragonscale cells will generate power. Once completed, the roofs will generate about 40% of the buildings' energy needs. 

The panels were designed in a standard size with standard mounts to be more manufacturable. According to a report in Grist, SunStyle has installed similar arrays on some 500 roofs in Europe, but only two in the U.S. -- a bank in Illinois and an architect's home in New York. 

The company now has plans to launch more affordable dragonscales targeted at homeowners looking to replace or install a new roof. The cost should be on par with a new roof with solar panels; however, dragonscales can also cover eaves and ridges. 

DARPA Targets New Night Vision

Last week, DARPA selected 10 teams to work on its Enhanced Night Vision in Eyeglass Form (ENVision) program. The idea is to create a pair of night vision glasses that are the same size and weight as regular eyeglasses. 

Night vision systems are bulky and weigh heavily on a soldier's neck. Over time, they can lead to chronic injuries. DARPA doesn't just want to make them lighter, but also give them a wider field of view and the ability to use infrared bands to see through obstacles. 

The teams are a mix of eyewear companies, Department of Defense staples like Raytheon and Physical Sciences, and universities.

Dog Robots Get Better at Walking 

We've mentioned before how robots could soon be used in remote and hazardous environments that are challenging, if not impossible, for humans to traverse, like looking for survivors after a natural disaster or industrial accident. While previous tech has been promising, one of the biggest obstacles has been the robots' ability to quickly assess and navigate the terrain. 

One of the biggest challenges is how the robots see the ground and adjust their gait to remain upright and stable. For example, robots have trouble walking through snow, vegetation or water because they are highly reflective surfaces, and the robot doesn't know what it's stepping on or the true height of the ground. They also struggle with depth perception in foggy, dusty or dimly lit areas. If they're going to be useful after a disaster, they're going to be in some dusty and dimly lit areas. Until now, robots have had to step down and feel the terrain before adjusting their gait, which really slows them down. 

A team of researchers at ETH Zurich in Switzerland has developed a solution. It's a new perceptive locomotion controller strapped it to an ANYmal, which, for better or worse, is often referred to as a robot dog. Then, they tested it in multiple scenarios that would typically mean trouble for robot dogs. 

Perhaps the most impressive was walking through high snow, but the ANYmal also climbed stairs in fog, completed hiking paths without falls or human intervention, and was even used to study monkey/robot interaction in South Africa. The importance of robot/animal interaction aside, the robot performed in each environment using the same settings and didn't experience a single failure. 

Team CERBERUS also used the locomotion controller for DARPA's Subterranean Challenge. The robot operated autonomously underground without a single fall and helped the team net $2 million in prize money. 

Finally, like many a robot before it, and many to come, they had to test it for human intervention. So the researchers tried to throw it off course by pushing it and dragging it on a leash. It's funny and cruel, in a robot-feelings sort of way, but, remarkably, this ANYmal stays on its feet. 

More in Video