Autonomous Vehicles That Don't Crash
As autonomous vehicles become a reality, we need algorithms that help them drive without crashing into one another, or quitting altogether when security protocols are breached. It's one of the biggest challenges for roboticists who create behaviors for teams or swarms of robots.
Each robot is given an invisible bubble that cannot be breached, but when you have enough robots working together, they shut down because bubbles are breached.
Well, a team out of Georgia Tech's Institute of Robotics and Intelligent Machines has created new algorithms that allow any number of robots to come within inches of one another, without colliding.
Their bots are using "a set of safe states and barrier certificates to ensure each stays in its own safe set throughout the entire maneuver." Essentially, they shrunk their robots' bubbles.
In a recent demo, the researchers proved the theory when swarm of as many as eight robots worked in close quarters with one another without shutting down.
Now, is this ready for the freeway, or gridlock traffic downtown? Maybe not, but it's a step in the right direction - and it could even be considered for next generation air traffic control, as we find a way to more safely pack those planes flying our friendly skies together.
Van with a Vision
The Vision Van from Mercedes-Benz Vans is a new concept vehicle that was developed with drone-delivery startup Matternet. As you can imagine, it was built to meet the need should the delivery-by-drone business ever take off.
The concept van is sort of a hybrid FedEx truck with a built in drone helipad. As the delivery driver, who remains human at least in this concept, delivers packages in-person, a drone fleet delivers packages to different customers in the area.
According to Daimler, the vehicle merges a number of innovative technologies for last-mile delivery operations. It features a fully automated cargo space, integrated drones for autonomous air deliveries and a state-of-the-art joystick control.
Powered by a 75 kW electric drive system with a range of more than 160 miles, the van would run cleaner, and be virtually silent, which would make stealthy late deliveries in residential areas possible to suite same-day delivery.
Malicious Spies Steal 3D-Printed Designs
In March, we saw how a team of researchers at the University of California, Irvine found a way to reverse engineer 3D-printed designs by simply recording the sound the process makes while being built in a 3D printer.
Well, a new study out of the University of Buffalo has proven 3D printers even more vulnerable to malicious spies.
Researchers programmed a smartphone's built-in sensors to measure the electromagnetic energy and acoustic waves that emanate from the 3D printer. The sensors can infer the print nozzles location as it builds the object.
With a smartphone 20 cm away, they were able to gather enough data to replicate simple designs within 94% accuracy, and more complicated designs, like an automotive part or medical device, within >90% accuracy. It's an improvement of the 90-percent accurate knockoffs they were making at UC-Irvine.
So what can we do to keep disgruntled employees and industrial spies for stealing our stuff? It could be as simple as keeping devices further away from the printers - as the method was only 66-percent accurate when the phone was 40 cm away from the printer. Signal jammers or even programming the printer to operate at different speeds could help.
This is Engineering By Design.