BrainChip Gives the Green Light to Automotive Innovation

For more than 90 years, the car radio has been a ubiquitous accessory found in automobiles worldwide.  Whether catching up on news, weather or traffic, or rocking out to the oldies or Top 40 songs, its hard to imagine riding in silence on trips to the grocery store or cross country.  And while we take these audio devices for granted today, getting them integrated into cars wasn’t an overnight success.

Radio technology was first considered for cars as early as 1904 but speech and music transmission wasn’t even feasible for another 5 years or so.  It would take another decade before the technology of vacuum tubes would progress to the point where car radios could exist but issues of voltage, electric interference from engines, heat dissipation from tubes and the size of components were still impediments to its success.  When the first after-market radios were sold, the cost was so prohibitive, only the wealthy could even consider them.

The Gavin brothers came up with a way to break up the components – with the receiver, controller, speakers, batteries and antennas installed in different sections of the car and wired together to achieve the necessary radio reception to produce a stable sound experience.  The resulting company created from this success – Motorola – is still a leading telecommunications enterprise today.

In the subsequent years, hundreds of technologies have been integrated into the automobile. Volkswagen introduced the first vehicle with a computer controlled electronic fuel injection system in 1968.  The 1970s saw Japanese auto makers integrate circuits and microcontrollers into windshield wipers, electronic locks and engine controls.  Tire sensors, climate control, electronic brakes, blind spot detection systems and more have all transitioned from nice-to-haves to must-haves among car buyers.

Today, processors are being implemented in car systems leveraging AI algorithms to advance “Smart Cars” from Sci-Fi to reality.  And like the early attempts at successfully installing a working radio into a car’s dashboard, various applications have had varying results as manufacturers look to overcome limitations in technology for a more seamless experience.

Low Power, High Functionality

One of the biggest limitations that the automotive industry faces in this transition is the compute power required to make calculations in real time.  As the industry moves from simple mechanical machines to electronic systems actually controlling propulsion, braking, maintenance, safety, and in cabin automation there is an increasing need to find solutions that provide significant reductions on battery consumption to ensure that vehicles retain their maximum efficiency.

A limitation of EVs often overlooked by consumers (the limitation is well known to manufacturers) is that batteries degrade over time.  We see it with many electronic devices that need frequent charging – battery cells can become “leaky” with maximum charge capacity lessening the more a vehicle is driven.  Having dozens or hundreds of “beefy” electronic processors can hasten the decline on a car’s electric and battery systems.  Implementing an ultra-low power, flexible, self-contained, event-based neural processor that is capable of inferencing and learning to support today’s most common neural networks can have a profound effect on the efficiency of vehicles and their effect on the environment.  These neural-network processors can extend battery life by more than 50+ miles per vehicle.  And with millions of cars on the road, this exponentially improves the need for charging and minimizes how many spent batteries are disposed of over time.

Many AI training modules require sensors to collect data and send it to the cloud for analysis.  While to human perception, this process is extremely fast, the level of latency required to calculate is far too long in real-world settings.  Object detection of stop signs or lights might be more readily accessible through Deep Learning algorithms but can these processors detect the difference between a rock and a plastic bag quick enough or with enough accuracy effectively?

All the Feels

A lot of the focus when discussing automotive sensors tend to center on visual-based sensors but the automobiles being produced right now can leverage multiple senses to determine problems.  In addition to cameras, there are sensors that can detect a smell that indicates oil is burning or that a cabin filter needs to be replaced.  Other sensors can feel or hear vibrations of various parts to diagnose whether a part has become lose or whether something is damaged.

When operating in a healthy state, a piece of machinery produces an expected vibration or sound.  As that equipment ages over time, those sounds or vibrations change. By the time humans can perceive these changes – something’s “rattling” or there’s a constant humming where there wasn’t any sound before – an edge AI processor can sense an impeding problem through real-time analysis of sensor data.  This early detection can help solve problems at a much earlier stage and reduce deterioration in real-time, providing invaluable progress towards complete safety and reducing maintenance costs.


The automotive industry is undergoing a tremendous transformation right now.  From government mandates requiring manufacturers to move away from gas-powered vehicles to businesses looking to minimize labor costs by adopting autonomous vehicles, the future begins now.

AI and Deep Learning tools are already being deployed in the industry in an attempt to make these cars smarter and safer.  But much like implementing radios into cars in the early 20th Century, there are issues and complexities that are complicating these efforts.   By moving AI out of the data center to the location where data is created, a lot of these problems can be averted.

BrainChip’s Akida AI neural processor is an event-based technology that is inherently lower power when compared to conventional neural network processors. By allowing incremental learning and high-speed inferencing, Akida overcomes current technology barriers through a high-performance, low-cost, very efficient low-power solution. It’s the kind of solution that is music to the ears of automotive manufactures and OEM’s looking to bring their ideas to life. 1000 miles on a single charge here we come!

BrainChip: This is our Mission

To learn more about how BrainChip’s approach to automotive sensor applications gives the green light to innovation, visit