We're still sifting through the more than 1,200 presentations at IROS 2017, IEEE's massive intelligent robots conference held last month in Vancouver. This week we found some terrifying gems: surgical robots that snake up the nose, puncture the breast, and suction intestinal tissue with motions so jarring they will make any patient glad (or longing) to be passed out during the procedure. We previously highlighted 20 of our favorite videos from the conference. Now, with Halloween approaching, we give you: the five most gruesome (maniacal laugh). Sadistically named the "Stormram," this robot punctures the breast in one slow, ominous motion, to extract a tissue sample.
The U.S. Federal Aviation Administration's Small UAS Rule (also known as Part 107) has provisions to obtaining waivers to the usual requirements for flying drones in the United States. For example, you're not generally allowed to fly drones at night, although the FAA has granted quite a few waivers allowing flight after dark. But another rule is that you can't fly drones over people who are not part of your operations, and until about a week ago, the FAA hadn't waived that rule for anybody. Now it has, for CNN. The FAA is allowing the cable news network to use a drone to obtain video over uninvolved people, even crowds assembled at places like sporting events.
There are many reasons to admire the octopus, including its ability to instantly pop up tiny protrusions of various shapes from its skin to match the texture of its background. This technique, combined with other camouflage tricks such as changing its color, allow an octopus to blend into to almost anything--even boats. Those protrusions, called dermal papillae, were the bio-inspiration behind a new elastic material that can morph into various shapes, and could provide a shape-shifting surface for soft robots. Researchers from Cornell University in New York and the Marine Biological Laboratory in Massachusetts decided to build a material based on muscle groups that control papillae along the surface of an octopus tentacle. The material consists of a fiber mesh that simulates an octopus's erector muscles, which contract to squeeze a protrusion into shape.
In the 1970 sci-fi thriller Colossus: The Forbin Project, a computer designed to control the United States' nuclear weapons is switched on, and immediately discovers the existence of a Soviet counterpart. The two machines have become one, and it has mankind by the throat. Development work takes a lot longer than that. Today DeepMind, a London-based subsidiary of Google, announced that it has developed a machine that plays the ancient Chinese game of Go much better than its predecessor, AlphaGo, which last year beat Lee Sedol, a world-class player, in Seoul. The earlier program was trained for months on a massive database of master games and got plenty of pointers--training wheels, as it were--from its human creators.
In 2015, two American engineers, Gui Cavalcanti and Matt Oehrlein, set out to build a giant human-piloted combat robot called Mk. II MegaBot that could drive on tank tracks and fire 3-pound projectiles. The robot was pretty cool, they thought, but who would they fight? So they decided to challenged the only other giant piloted robot in the world to a duel. That robot was a 4-metric-ton mech known as Kurata and built by Suidobashi Heavy Industry in Japan. The Japanese accepted the challenge.
Of all the legged robots built in labs all over the world, few inspire more awe and reverence than Boston Dynamics' quadrupeds. Chinese roboticist Xing Wang has long been a fan of BigDog, AlphaDog, Spot, SpotMini, and other robots that Boston Dynamics has famously introduced over the years. "Marc Raibert … is my idol," Wang once told us about the founder and president of Boston Dynamics. Now Wang, with funding from a Chinese angel investor, has founded his own robotics company, called Unitree Robotics and based in Hangzhou, outside Shanghai. Wang says his plan is making legged robots as popular and affordable as smartphones and drones.
Now Kernbaum is back with another ingenious transmission design: an ultra-compact, infinitely variable transmission based on a novel nested-pulley configuration. In an infinitely variable transmission, which is a specific kind of continuously variable transmission, the transmission ratio includes a zero point that can be approached from either a positive side or a negative side. In other words, a constant input, like an electric motor turning the same direction at the same speed, can be converted to an output that's turning faster, turning slower, turning in the opposite direction, or not turning at all (in this "geared neutral" mode, you'd need infinite input revolutions to cause one output revolution, hence the name "infinitely variable transmission"). If you can't quite understand how it works from the video (and even Kernbaum admits that it's difficult to visualize), read the explanation here: Inception Drive: A Compact, Infinitely Variable Transmission for Robotics
Deep reinforcement learning (DRL) provides a model-agnostic approach to control complex dynamical systems, but has not been shown to scale to high-dimensional dexterous manipulation. Furthermore, deployment of DRL on physical systems remains challenging due to sample inefficiency. In this work, we show that model-free DRL with natural policy gradients can effectively scale up to complex manipulation tasks with a high-dimensional 24-DoF hand, and solve them from scratch in simulated experiments. We demonstrate successful policies for multiple complex tasks: object relocation, in-hand manipulation, tool use, and dooropening.
In a recent set of experiments, STAR's inventors showed that it makes more precise cuts than expert surgeons, and damages less of the surrounding flesh. Before they tested the system against human surgeons, STAR first had to prove its ability to make precise cuts in these three types of irregular soft tissue, which can resist a cutting tool and then give way abruptly, causing the tool to make inaccurate cuts. "After registering the markers, the robot autonomously performs the tumor cutting," Krieger says. In the final stage of the experiment, the researchers had STAR cut a fake tumor (made of clay) out of a piece of pig fat; a thin piece of tissue was placed above the fake tumor to make it harder for the robot to see its target.
Within 24 hours, a team of CMU roboticists had packed their gear and headed out to the disaster site. We spoke with Matt Travers, who was on the ground in Mexico City operating the robots, along with Howie Choset, who heads CMU's Biorobotics Lab where the snake robots are developed, about their experience with using robots in a real disaster and how, although no survivors were found during the rescue missions they assisted with, they learned an enormous amount being on-site. Travers: We showed the Red Cross workers what the robot was capable of when we first got there, but the difficulty wasn't in convincing people that the robot could be useful--it was difficult for anyone to get to the sites, no matter what their technology or specialization was. To put this in perspective, we were with the same team of Red Cross workers all day Friday, all day Saturday, and all day Sunday, and they only brought out their plungers and microphones twice, while the robot was used once.