Presented by:
One of Intel’s Nahuku boards, each of which contains 8 to 32 Intel Loihi neuromorphic chips, shown here interfaced to an Intel Arria 10 FPGA development kit. Intel’s latest neuromorphic system, Poihoiki Beach, annuounced in July 2019, is made up of multiple Nahuku boards and contains 64 Loihi chips. Pohoiki Beach was introduced in July 2019. (Credit: Tim Herman/Intel Corporation)
SANTA CLARA, CALIF. – Intel today announced that its 8 million-neuron neuromorphic system that comprises 64 Loihi research chips – code-named Pohoiki Beach – is now available to the broader research community. The brain-inspired research chip enables users to process information up to 1,000 times faster and 10,000 times more efficiently than CPUs for specialized applications that include sparse coding, graph search and constraint-satisfaction problems, Intel said.
Rich Uhlig, managing director of Intel Labs, holds one of Intel’s Nahuku boards, each of which contains 8 to 32 Intel Loihi neuromorphic chips. (Credit: Tim Herman/Intel Corporation)
More importantly for the robotics community, Intel said researchers “can now efficiently scale up novel neural-inspired algorithms” that include simultaneous localization and mapping (SLAM) and path planning. These algorithms are key to autonomous vehicle development.
“We are impressed with the early results demonstrated as we scale Loihi to create more powerful neuromorphic systems,” said Rich Uhlig, managing director of Intel Labs. “Pohoiki Beach will now be available to more than 60 ecosystem partners, who will use this specialized system to solve complex, compute-intensive problems.”
Special architecture for speed gains
Intel said that continuing the gains in power and performance enabled by Moore’s Law “will require more than continued process node scaling. As new complex workloads become the norm, there is a growing need for specialized architectures designed for specific applications.”
The Pohoiki Beach neuromorphic system demonstrates the benefits of this specialized architecture, including some of the computational problems for IoT and autonomous devices to support, the company added. With this system, Intel said companies can expect to realize orders of magnitude gains in speed and efficiency.
“Loihi allowed us to realize a spiking neural network that imitates the brain’s underlying neural representations and behavior,” said Konstantinos Michmizos, a professor at Rutgers University. “The SLAM solution emerged as a property of the network’s structure. We benchmarked the Loihi-run network and found it to be equally accurate while consuming 100 times less energy than a widely used CPU-run SLAM method for mobile robots.” His lab’s work on SLAM will be presented at the International Conference on Intelligent Robots and Systems (IROS) in November.
Researchers at the 2019 Telluride Neuromorphic Cognition Engineering Workshop are working to automate Western Sydney University’s foosball table under Loihi control, operating on visual input from event-based cameras. Foosball offers an excellent test for rapid closed-loop sensing, planning and control algorithms, a sweet spot for neuromorphic technology. (Credit: Sumit Bam Shrestha)
Researchers at the Telluride Neuromorphic Cognition Engineering Workshop this week also cited examples of how they were using Loihi systems to solve challenges. Projects included providing adaptation capabilities to a prosthetic leg, object tracking with emerging event-based cameras, automating a foosball table with neuromorphic sensing and control, controlling a linear inverted pendulum, and inferring tactile input to the electronic skin of an iCub robot.
The Loihi chip was first introduced in 2017, and today’s announcement of Pohoiki Beach accelerates the development by opening up the capacity and scale to Intel’s research partners, the company said. Later this year, Intel said it will introduce a larger Loihi system named Pohoiki Springs, which builds on the Pohoiki Beach architecture “to deliver an unprecedented level of performance and efficiency for scaled-up neuromorphic workloads.”
