BMW will study the possibility of introducing neuromorphic chips into cars. These devices resemble the human brain in architecture, and neural networks created on their basis work hundreds of times faster than conventional ones. BMW’s partner in this area will be the Chinese company SynSense. The company’s portfolio includes the Speck neuro processor with an event-based camera, which can be used in pedestrian recognition and traffic analysis systems. It is not yet known when the neuromorphic processors will appear in production cars.
Neuromorphic processors are complex devices, their architecture resembling the human brain. They contain neurons and analogs of synapses, organized into pulse (spike) neural networks. Chips support “deep learning” methods, do not require an external database and have low power consumption, and have a high speed of operation. But examples of their application in cars can be counted on the fingers of one hand.
Loihi 2 is Intel’s second-generation neuromorphic chip. It uses new classes of algorithms and is more productive and energy-efficient.
For example, in 2020 Mercedes-Benz announced the beginning of cooperation with Intel’s research division, Intel Labs. The Germans were interested in neuromorphic computing technology, which would significantly improve voice and gesture recognition systems. According to Intel itself, the chip it created consumed 1,000 times less power and was 0.2 seconds faster than a conventional graphics processor.
This winter, neuromorphic processors made their first appearance on a car, the Vision EQXX concept car. The chips and necessary software were provided by BrainChip, a Californian company, and they were responsible for processing voice and gestures. Although the current level of technology allows increasing the accuracy of determining the phrase “Hey Mercedes” by 5-10 times, the area of neuromorphic computing is still in its infancy, and it is still too early to talk about its commercialization.
Meanwhile, BMW has decided to do similar experiments – but Bavarians will use neuromorphic processors for machine vision systems. This is the Speck computing chip from China’s SynSense, which combines an image processor with a convolutional neural network and an event-based camera on a single chip (SoC). The latter is an asynchronous sensor that responds to local changes in brightness that exceed specified thresholds.
The event camera captures only moving objects, i.e., information is transmitted only from active pixels. Therefore, the SoC system consumes one milliwatt of electricity, and the delay time is 0.005-0.01 seconds. In theory, such devices can be used for object detection, face recognition, pedestrian tracking, and traffic analysis. However, it is not yet known in what capacity neuromorphic processors will be used in BMW.
Source: SynSense