Not many would possibly remember a hardware bug that affected the floating-point unit (FPU) of the early Intel Pentium processors. This was in the early 1990s. The bug led the processor to return incorrect binary floating point results, when dividing certain pairs of high-precision numbers.
Intel recalled the defective processors in December 1994, in what was the first full recall of a computer chip. A small company then, it cost Intel half a billion dollars in recalls. As a result though, the whole industry started to focus on how to verify and validate microprocessors, and different methods were developed.
Three Israeli engineers were right in the middle of that industry, developing solutions to address challenges of solving verification and validation problems for sophisticated microprocessors. These were complex problems in the chip design industry, and they had to ensure they function correctly.
A decade later, two of these engineers – Yoav Hollander and Ziv Binyamini – worked together in a startup that offered solutions to the rest of the chip design industry. This company was later acquired by Cadence, an American multinational computational software company.
Around 2016-17, Hollander and Binyamini, together with the third engineer Gil Amid realised that the solutions they developed were also applicable to verify the complex software and hardware systems of autonomous vehicles (AVs).
The trio figured out that the same approach applied to solve and verify 10 billion transistors in a semiconductor chip could be modified and applied to how one can verify a self-driving car that meets an infinite set of situation and scenarios in real life.
One thing led to another, and Foretellix was formed with Yoav Hollander as the Founder & CTO, Ziv Binyamini as the CEO & Co-founder and Gil Amid as the Chief Regulatory Affairs Officer, VP Operations & Co-founder. The idea was to enable the digital transformation of safety-critical automotive verification. The startup raised some initial seed funding, and began to create a solution for the self-driving car industry.
Foretellix’s Solution?
Foretellix’s USP lies in a “very specific software language” that can be used to write scenarios. It was almost three decades back that Hollander had invented such a language for chip design. He has now refined this language and tuned it completely towards the automotive industry.
In a recent interaction with Mobility Outlook, Amid said the language, which would earlier describe “how you feed two plus two into a complex microprocessor”, can now describe different events happening in the road, including “how a car is driving, and how is someone cutting into its path, etc”.
Once you have such a language and scenarios, he added, the rest is simple. “You need a computer programme that will read this language, understand it, and then create the right situation inside the simulator,” he explained.
One of the biggest challenges of a self-driving vehicle is that one can’t be sure if it is safe or has been tested enough. Foretellix’s solution aims to enable and improve “measurable safety” of ADAS (Advanced Driver Assistance Systems) and autonomous vehicles. It aims to ensure the autonomous vehicles behave properly under all possible driving conditions.
The current level of testing on AVs and the numerous possible situations are insufficient, said Amid. The Foretellix solution enables companies to undertake very intensive and thorough testing using simulation. It develops intelligent automation and analytic tools to orchestrate and monitor hundreds of millions of driving scenarios, including edge cases.
Also Read:Magna Develops, Pilots Autonomous, On-Road Last-Mile Delivery Solution
Different Scenarios
“We are not offering any simulation capabilities, but rather hooking up to existing simulation capabilities by other developers and OEMs,” said Amid, adding “Foretellix is a layer on top of the simulator”. Essentially, Foretellix injects into the simulator different scenarios and measures the response of the self-driving cars to these scenarios.
“Our job is to challenge software of the self-driving car and see whether it reacts correctly or not,” said Amid. Explaining further, he said the Foretellix solution is being used during the development and certification phases of the vehicle. It is not a software that is sitting on the vehicle while it is in operation.
While simulating the various situations is one part of the puzzle, the other – and perhaps the more critical – is to measure the behaviours and the responses. For example, are you keeping safety distance or are you getting into a collision?
Amid said the Foretellix solution reads such scenarios, simulates them, and measures & accumulates the results from all such scenarios. This helps in understanding the behaviour of the self-driving vehicle.
Building Standards
As a young startup from Israel with a sophisticated software language addressing the challenges of autonomous vehicles, finding its feet in the strictly standards- and regulations-driven automotive industry wasn’t going to be easy. Its solution had to have a huge influence on the industry.
Early in its journey, the company understood that with its proprietary language, winning the confidence from large automotive players was going to be tough. “What are the chances that a giant like BMW would rely on a proprietary language coming from Israel?” reasoned Amid.
Their search led them to the Association for Standardization of Automation and Measuring Systems (ASAM), a standardisation organisation based in Höhenkirchen, Germany with the primary objective of standardising development and test systems for the automotive industry. Foretellix was able to convince ASAM and its members that its language was the right approach for complex high scale validation.
Project Leader
ASAM launched a project in 2019 to develop a standard called OpenSCENARIO, and Amid took over as its project leader. In July 2022, ASAM’s OpenSCENARIO 2.0 standard was released, with Open M-SDL (Measurable Scenario Description Language) as its foundation.
M-SDL is an open, human-readable, high-level language that simplifies the capture, reuse, and sharing of scenarios and easily specifies any mix of scenarios and operating conditions to identify previously unknowns and edge cases.
The language is used by tool vendors, suppliers and developers in the ADAS & AV development space, as it allows to monitor and measure the coverage of autonomous functionality critical to proving ADAS & AV safety, independently of test and testing platforms. M-SDL was chosen as the concept demonstrator and later adopted as the basis for ASAM’s OpenSCENARIO 2.0, said Amid.
“The specification of the language is available free of charge to anyone who would like to download it from ASAM,” he said. Interestingly, following ASAM’s OpenSCENARIO 2.0 standard publication announcement, Foretellix suspended the development of M-SDL to support the industry transition to OSC2. There are no plans to release other versions of the M-SDL.
Meanwhile, ASAM is working on a project to accelerate migration from the previous generation of the language, which Amid said was “very naïve and insufficient”.
Snowball Effect
The Tel Aviv-based startup is now starting to see the snowball roll. In its rather short journey, it has acquired several key global customers, including the likes of Volvo, Denso, Nvidia, Valeo, Mobileye, Nio, AWS and IPG Automotive, among others. There are many other partnerships that Amid said are too premature to talk about.
In January this year, the company announced a multi-year strategic collaboration with Torc Robotics, an independent subsidiary of Daimler Truck AG for automated driving systems (ADS). The collaboration delivers to Torc the best-in-class virtual verification & validation solution, capable of testing the millions of driving scenarios required to ensure the safe deployment of Level-4 autonomous trucks.
Earlier in November last year, Foretellix and Cognata had joined hands to bring together Cognata’s developers’ simulation engine, Enterprise, with Foretellix’s verification and validation platform, Foretify. The joint solution helps ensure the safety of ADAS and AV systems, reduces development costs, and improves time to market, claimed a release the company had issued at that time.
Another key partnership for Foretellix is the one with Volvo Autonomous Solutions. This partnership aims to jointly create a coverage-driven verification solution for autonomous driving that operates both on public roads and in restricted areas. Based on its experience working with Foretellix, Volvo Group Venture Capital AB later decided to invest in the Israeli company.
In India, Foretellix has a partnership with Bengaluru-based Vayavya Labs that offers embedded software tools and solutions. For now, Foretellix is outsourcing activities to Vayavya, and Amid believes they will be developing some of their own solutions as well. There are some discussions with other key players in the Indian market, confirmed Amid, but he refrained from talking about them due to non-disclosure agreements.
Future Growth Drivers
In addition to its current offerings to the market, Foretellix is developing other products, including one that reads video logs and recordings, and translates them into scenarios. This offers two benefits, said Amid – “One, you know what you tested during real driving, and second, you immediately can use it for more verification”.
As far as competition is concerned, there are other companies that are writing software to process the language, he continued. “The language is not proprietary. The specification of the language is now a standard, which is available for everyone. But we are the most advanced, and of course, first to the market because we knew about the language early enough. We invented it,” said Amin.
He picked Applied Intuition, a US-based competitor in particular. Although they started with a different approach, they’re now trying to adapt OpenSCENARIO 2.0, he said.
“Our belief is that we offer technologically superior solutions that understand the language, utilise it to the best, and deliver the best productivity and quality improvement,” Amid said.
Safety At The Core
At the last Consumer Electronics Show (CES) in Las Vegas, Foretellix announced the launch of Foretify, a Safety-Driven Verification & Validation (SDV) Platform, which essentially are a full set of steps based on its software. And since it is safety driven validation, “we call it SDV, not to be confused with software-defined vehicles,” joked Amid.
A delighted Amid said the momentum is building for Foretellix, and the language is catching up. Of course, it has the first mover advantage, but the challenge of the autonomous world is too large and complicated still, and it will need to continue to develop technological solutions and also scale up its business.
To that extent, the company at the start of this month announced it has raised $43 million in the first closing of its Series C funding round led by 83North, bringing its total raised capital to over $93 million. Foretellix will use the funding to accelerate development of its expanding product portfolio and fuel expansion across new geographies.
Woven Capital, the growth fund of Toyota, and NVIDIA joined the financing round, along with Artofin VC and all major existing shareholders, including MoreTech, Nationwide, Volvo Group VC, and Jump Capital, participated as well.
Helmed by some of the veterans of the software industry, Foretellix’s vision of addressing the largest barrier for safe large-scale deployment of automated driving systems, is on a strong footing. For now, Level-4 and Level-5 autonomous mobility might continue to be a distant dream, but with innovations like the ones introduced by Foretellix and others in the industry, safe autonomy might not be too far.
Also Read:
Managing Cybersecurity Software At Scale Key To Future Mobility
Toyota Blends 'Autonomous' With 'Mobility-as-a-Service' In Sienna Autono-MaaS