This page contains a Flash digital edition of a book.
TECH TALK


SMARTER CHIPS — THE RISE OF NEUROMORPHIC CHIPS (AKA SKYNET)


AeroVironment has designed and built a tiny new quadcopter with partial funding coming from DARPA’s neuromorphic SyNAPSE project. This drone was six inches square, 1.5 inches high, and weighed 93 grams including the battery. The neuromorphic chip that controlled this drone used only 50 milliwatts of power (not much) and was able to process information from optical, infrared and ultrasound sensors as it flew between three rooms, essentially acting like some living beings. Why am I mentioning this as an electronics industry


innovation and not an aerospace innovation? Quite simply, this copter could not have been built without the use of a brain-like neuromorphic chip. These chips differ from the traditional chips of today that are based upon the ‘von Neumann architecture’ (this is your computer science history lesson for the day), which is essentially little- changed at its core of the years. A majority of computers today use this von Neumann architecture, which transports data back and forth between a central processor unit and memory chips in linear sequences of calculations. This is simpler than how a human brain processes data. The von Neumann method is good at running applications code or sorting through sets of numbers or data, but is not good at handling and understanding visual and audio information. This is where the new computing architecture based upon neuromorphic chip architecture provides a vast improvement, as shown in the table from the MIT Technology Review. According to DARPA, “The vision for the Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) program is to develop low-power electronic neuromorphic computers that scale to biological levels. Current computers are limited by the amount of power required to process large volumes of data. In contrast, biological neural systems, such as the brain, process large volumes of information in complex ways while consuming very little power. Power savings are achieved in neural


systems by the sparse utilizations of hardware resources in time and space. Since many real-world problems are power limited and must process large volumes of data, neuromorphic computers have significant promise.” There are several companies experimenting with


building these chips. The one used for the quadcopter test was supplied by the HRL Laboratories (aka, the Howard Hughes Research Labs, which is owned jointly by GM and Boeing). Qualcomm is also betting big on this as its existing patent portfolio is under attack and it could use something new. (China has threatened to fine it ~$1B for apparently overcharging on its licensing fees on phone chipsets that use its patents in Chinese firms.) If these new chips are able to learn faster, process various types of data faster and use less power, Qualcomm and others (like IBM) could control the guts of the next generation of mobile computing for years to come, as well as for other application areas (like transportation, industrial, medical, etc.). Essentially, once the technical challenges are overcome in the much larger technology market for how chips are deployed, you can expect aerospace firms to rapidly incorporate neuromorphic-based CPUs into systems that need to acquire various external data and process them quickly (like navigation systems, weather tracking, etc.). Note that not all aviation-related systems would need to make use of such chips — only those that rely more on acquiring unstructured data and less on processing software applications to accomplish some function. In this case, there is no comparison to any effort in the


aerospace market, since aircraft and aviation support systems will be consumers of this chip. In fact, these neuromorphic chips cannot come quickly enough since they promise to increase the quality of many airborne, ground and satellite systems and the resulting data they will be able to provide with faster processing times and reduced power consumption needs. In fact, avionics, ground-based systems and airborne sensors with such capabilities might be a saving grace for NextGen and put off its obsolescence.


Processing Powers What they do well


Neuromorphic chips Detect and predict patterns in complex data, using relatively little electricity


Traditional chips (von Neumann architecture)


Reliably make precise calculations


What they’re good for


Applications that are rich in visual or auditory data and that require a machine to adjust its behavior as it interacts with the world


Anything that can be reduced to a numerical problem, although more complex problems require substantial amounts of power


03 2015 32


DOMmagazine


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56  |  Page 57  |  Page 58  |  Page 59  |  Page 60  |  Page 61  |  Page 62  |  Page 63  |  Page 64  |  Page 65  |  Page 66  |  Page 67  |  Page 68