Trailblazing Self-Driving Into the Tangible Present

While most are familiar with the Israeli giant Mobileye which has been around since 1999, it is Autobrains that is perceived by many major automakers to be the next big thing in the Advanced Driver-Assisted Systems (ADAS) industry. The company is backed by investments from auto giants such as Toyota, BMW, Continental, Knorr-Bremse and others, after they assessed its new impressive technology.

Autobrains, which was founded in 2019 as a spinoff of the Cortica group (a leading AI technology powerhouse for autonomous platforms operating in several verticals), uses groundbreaking technology backed by more than 250 patents and ten years of cutting-edge research. “Our technology is based on self-learning AI which enables cars to learn, collaborate and interact with the world like humans, without human supervision. That’s the essence of what makes Autobrains’ novel AI capable of overcoming the hardest challenges where current systems are plagued by barriers,” explains Autobrains founder and CEO Igal Raichelgauz.

The underlying technology was developed at Cortica when Raichelgauz and Cortica’s two other founders, Karina Odinaev and Prof. Josh Zeevi, met when Igal and Karina were doctoral students at the Technion-Israel Institute of Technology. Their research project led to a series of breakthrough discoveries into how the mammal cortex deciphers visual environments. The three combined their expertise in Neuroscience, Electrical Engineering and Computer Science to hack into live rat brain tissue and “encode” digital images. They later founded Cortica to develop autonomous AI applications by mimicking these natural biological processes. Prior to founding Autobrains, Raichelgauz served as Cortica’s CEO for more than ten years and remains its Chairperson.

Superior performance

Today, almost all newly purchased cars are equipped with some level of ADAS functions, designed to eliminate the human error factor in driving by processing the signals from the road scenery received via cameras and other sensors, alerting drivers to obstacles detected on the road and even enabling semi-autonomous driving, as made famous by Tesla. In the future, vehicles will be increasingly autonomous. The rapidly growing ADAS market is projected to reach approximately USD 75 billion in annual global sales by 2030.

The ADAS developed by the Israeli company Mobileye has generally been considered the industry leader. However, it’s been reported that Mobileye and others have several shortcomings – most notably problems with biases and errors which lead to an inability to cope well with edge cases, as well as high power and computing requirements, and a problematic gap between perception to decision.

All the while, Autobrains has been quietly developing an ADAS system at accelerated pace that is not only significantly safer than competing solutions but is also dramatically less expensive.

After being tested on millions of miles by top automakers, Autobrains has proven itself to be safer than the competition. In particular, it demonstrated superior performance in edge cases, including bad weather conditions and incidents where pedestrians are hard to discern, to name a few. According to Raichelgauz, Autobrains would have been able to prevent recent accidents reported on the news where cars using other technologies faced edge case scenarios.

Along with improved safety, Autobrains’ additional advantage is the significantly lower costs that it offers its customers. By reducing reliance on labeled data, Autobrains’ technology requires far less computing power and can be produced at a fraction of the cost of current deep learning systems on the market (up to 40% lower).

“At Autobrains we aim to reach the industry goal of contributing to zero accidents – without high computing costs and manual processing that makes this unfeasible. We are disrupting how the automotive industry approaches autonomous car technology with a technology which takes it closer than ever to the human way of driving perception,” Raichelgauz explains.

How does it work?

Autobrains’ self-learning AI operates in a fundamentally different way from traditional deep learning systems. Autobrains’ signature-based approach simulates how the human brain functions, shifting from labels to generic representations. The sensory information is indexed by compressed, generic neural responses and later used for various semantic tasks. This lean approach is highly efficient, and more robust in edge cases than traditional approaches. With self-learning replacing supervised learning, neural networks are not reliant on extensive manually labeled data and it does not require the massive brute force typical of supervised learning AI – so the learning process is much quicker and immune against human error.

Together, signatures and self-learning mimic the human brain, and create a smart and powerful engine that not only classifies but comprehends the road environment. The technology is therefore as close as possible to how humans learn and react to their environment. In fact, using this novel self-learning technology, cars can learn to react to their surroundings in a manner that mimics human learning without human supervision. Unlike supervised artificial intelligence, Autobrains’ system learns from the surrounding environment in real-time and updates its databases accordingly.

Autobrains is set to solve the problem of autonomous driving and deliver safe and convenient ADAS functions along the way, based on the same underlying technology. Today, Autobrains’ software stacks support a full spectrum of functions – from entry-level through L2++, and on the way to full autonomy, ushering in the much-expected technological trajectory from ADAS to AV (autonomous vehicles).

Regulations currently mandate that new vehicles be equipped with safety features, including basics such as electronic stability control, anti-lock brakes, blind spot information systems, lane departure warning, adaptive cruise control and traction control. Some of the more advanced capabilities enable semi-autonomous driving: “school zone assist” helps drivers focus on pedestrians, “parking pilot” lets the car self-park, and “offroad profile” overcomes challenging road conditions.

Remarkable growth

Autobrains has been growing at a remarkable rate. Only two years ago, there were 35 people working for the start-up at its headquarters in Tel Aviv; today there are 120 employees, and that number is continuously growing, with new sites in Germany and the U.S. soon to be opened. The team includes engineers, neuroscientists, and machine learning experts, as well as professionals from academia and the automotive industry.

In late 2021, Autobrains announced its Series C financing of more than US$ 120 million, backed by some of the most prestigious and sophisticated financial and strategic investors in the automotive sector, including Temasek, a Singapore-based global investment firm, Knorr-Bremse, a leading automotive player, and VinFast, a fast-growing flagship Vietnamese car manufacturer, as well as existing investor BMW and long-term strategic partner Continental.

As a result of the technology’s successful testing and the global interest that it is attracting, Autobrains partnered with the most innovative companies in the automotive industry. To this end, Autobrains has already signed several important collaboration agreements with leading Tier-1s, automakers and Silicon players to work together to make autonomous driving safe and affordable for all.

“Along with our partners and vision to reverse engineer the human perception for autonomous driving, we will continue developing as the leading AI company for the safe and scalable transition to autonomous driving, and be the brain of every car,” concludes Igal Raichelgauz.

For more information, visit the website.