Toggle light / dark theme

AI has finally come full circle.

A new suite of algorithms by Google Brain can now design computer chips —those specifically tailored for running AI software —that vastly outperform those designed by human experts. And the system works in just a few hours, dramatically slashing the weeks-or months-long process that normally gums up digital innovation.

At the heart of these robotic chip designers is a type of machine learning called deep reinforcement learning. This family of algorithms, loosely based on the human brain’s workings, has triumphed over its biological neural inspirations in games such as Chess, Go, and nearly the entire Atari catalog.

Why do so many people get frustrated with their “high-tech” prostheses? Though sophisticated robotics allow for prosthetic joints that can do everything a human can and more, the way we control robotic machines right now doesn’t allow us to operate them as naturally as you would a biological hand. Most robotic prostheses are controlled via metal pads on the skin that indirectly measure muscle action and then make some assumptions to determine what the person wants to do. Whil… See More.


We plan to use MM to provide natural control over prosthetic limbs by leveraging the human body’s proprioception. When you wiggle one of your fingers, your brain senses muscle lengths, speeds, and forces, and it uses these to figure out the position of that finger. This is called body awareness, or proprioception. When someone receives an amputation, if their muscle connections are maintained with what is called the “AMI technique,” their brain still perceives muscle flexion as it relates to joint movement, as if their limb was still present. In other words, they are sensing movement of a phantom limb. To give an amputee intuitive control over a robotic prosthesis, we plan to directly measure the muscle lengths and speeds involved in this phantom limb experience and have the robot copy what the brain expects, so that the brain experiences awareness of the robot’s current state. We see this technique as an important next step in the embodiment of the prosthetic limb (the feeling that it is truly part of one’s body).

Notably, the tracking of magnetic beads is minimally invasive, not requiring wires to run through the skin boundary or electronics to be implanted inside the body, and these magnetic beads can be made safe to implant by coating them in a biocompatible material. In addition, for muscles that are close to the skin, MM can be performed with very high accuracy. We found that by increasing the number of compass sensors we used, we could track live muscle lengths close to the surface of the skin with better than millimeter accuracy, and we found that our measurements were consistent to within the width of a human hair (about 37 thousandths of a millimeter).

The concept of tracking magnets through human tissue is not a new concept. This is the first time, however, that magnets have been tracked at sufficiently high speed for intuitive, reflexive control of a prosthesis. To reach this sufficiently high tracking speed, we had to improve upon traditional magnet tracking algorithms; these improvements are outlined in our previous work on tracking multiple magnets with low time delay, which also describes how we can account for the earth’s magnetic field during portable muscle-length tracking. This is also the first time that a pair of magnets has been used as a distance sensor. MM extends the capabilities we currently have with wired-ultrasound-crystal distance sensing (sonomicrometry, SM) and tantalum-bead-based distance sensing via multiple-perspective X-ray video (fluoromicrometry, FM), enabling us to now wirelessly sense distances in the body while a person moves about in a natural environment.

The Ingenuity helicopter on Mars has now completed its 12th flight, where it acted as a scout, looking ahead for dangerous terrain for it’s partner in crime, the Perseverance rover.

The 4-pound autonomous rotocraft climbed over almost 10 meters (33 ft) high, and traveled a total of 450 meters (1,476 ft) in 169 seconds. It flew over the over an area dubbed the ‘South Seitah’ region of Mars, where Perseverance will explore.

“A dozen for the books!” said JPL on Twitter. “The Mars helicopter’s latest flight took us to the geological wonder that is the ‘South Seitah’ region.”

Program aims to provide physical systems with ability to adapt to unexpected events in real-time and effectively communicate system changes to human and AI operators.


Many complex, cyber-physical military systems are designed to last for decades but their expected functionality and capabilities will likely evolve over time, prompting a need for modifications and adaptation. High Mobility Multipurpose Wheeled Vehicles (HMMWV), for example, had a design life of 15 years, but are now undergoing modernization to extend the average age of the fleet to 37+ years. At design time, these systems are built to handle a range of expected operating environments and parameters. Adapting them is currently done in an improvisational manner – often involving custom-tailored aftermarket remedies, which are not always commonly available, require a skilled technician to install, and can take months or even years to procure. Further, as they evolve and are placed outside of their original design envelop these systems can fail unexpectedly or become unintentionally dangerous.

“Today, we start with exquisitely built control systems but then someone needs to add something or make a modification – all of which results in changes to the safe operating limits,” said DARPA program manager John-Francis Mergen. “These changes are done in a way that wasn’t anticipated – or more likely couldn’t have been anticipated – by the original designers. Knowing that military systems will undoubtedly need to be altered, we need greater adaptability.”

In response, DARPA developed the Learning Introspective Control (LINC) program. The program aims to develop machine learning (ML)-based introspection technologies that enable systems to adapt their control laws as they encounter uncertainty or unexpected events. The program also seeks to develop technologies to communicate these changes to a human or AI operator while retaining operator confidence and ensuring continuity of operations.