Although the complete 5G thing did not work out, Intel is still operating vigorously on its Loihi “neuromorphic” deep-learning processors, based on the human brain. It disclosed a new system, dubbed as Pohoiki Beach, composed of 64 Loihi processors and 8 Million supposed neurons. It is able to crunch AI algorithms almost 1,000 quicker and 10,000 times more competently versus normal CPUs for employment with prosthetic limbs, electronic robot skin, autonomous driving, and more.
The Pohoiki Beach system has various multiple Nahuku boards, which can be attached with developer’s kit (Arria 10 FPGA) by Intel. The Loihi processors are downloaded on a “Nahuku” board that has 8–32 Loihi processors.
Pohoiki Beach will be ideal for neural-akin tasks comprising path planning, sparse coding, and SLAM (simultaneous localization and mapping). In layman’s words, those are all methods employed for things such as indoor mapping for robots, autonomous driving, and efficient sensing systems. For example, Intel claimed that the boards are being employed to make specific kinds of prosthetic legs more flexible, reinforcing object tracking through efficient, new event cameras, offering tangible input to electronic skin of an iCub robot.
On a related note, a group of automotive firms and Intel have collaborated to make new rules for autonomous cars. The aim of the “Safety First Automated Driving” document is to set up a structure of universal security rules that all self-driving vehicles must follow. The standards address majorly how the sector must report and monitor safety protocols when operating and building autonomous vehicles.
Audi, Aptiv, BMW, Baidu, Daimler, Continental, Here Technologies, Fiat Chrysler Automobiles, Volkswagen, and Infineon were all comprised in authoring the paper, which set up 12 rules for autonomous cars. They comprise: operational design domain, safe operation, security, vehicle operator-initiated handover, vehicle-initiated handover, user responsibility, and so on.