Pioneering Automotive Safety: Provizio’s Path to ISO Certification in Machine Learning and AI

In the realm of automotive innovation, the future is driven by technology that not only aims for progress but embodies an unwavering commitment to safety. At the helm of this visionary journey is Provizio, an accident prevention technology company that uses advanced robotics, AI and proprietary radar designs to rapidly identify high-risk scenarios and take decisive action to avoid them. Led by Founder and CEO Barry Lunn, an entrepreneur and business leader in the imaging technology industry for over two decades, the Irish company stands as a beacon of transformation and an epitome of technology fused with safety.

Provizio’s collaboration with exida, a testament to their relentless pursuit of excellence, germinated in the fertile grounds of Future Mobility Campus Ireland. Here, on the cutting edge of innovation, they envisioned a seamless fusion of radar-based perception, Artificial Intelligence (AI), and Digital Signal Processing (DSP). Together with exida, they embarked on a journey to redefine safety standards, to pioneer the certification of machine learning algorithms in safety-critical applications.

The bedrock of this journey is Provizio’s 5D Radar technology, a marvel that ensures compliance with ISO 26262 and IEC 61508 certifications. Through unmatched resolution and extended range, this patented radar technology delivers a solid foundation for the integration of AI and DSP chains. The result? Superior perception, refined classification, and real-time scene understanding, all contributing to safer autonomous driving scenarios.

Yet, this transformation has not come without its fair share of challenges. Integrating AI into the safety-critical path demanded not just innovation but practicality. It required the collection of extensive training datasets, over 100,000 strong, reflecting diverse environmental conditions. The mission? To ensure cost-effective, widely adoptable technology, bridging the gap between safety, affordability, and efficiency.

As Provizio strides confidently towards certification, the focus remains unwavering: achieving a harmonious balance between safety, comfort, and cost. The ISO certification they pursue is more than a milestone; it is a testament to their unwavering dedication to road safety. Their aim is clear – setting the standard so others can follow suit, assuring end-users and OEM partners of the steadfast fundamentals that underpin their technology.

Looking ahead, the impact of this collaboration and the pursuit of ISO certification is colossal. Society’s trust in AI systems on the road is poised to elevate, akin to the trust placed in other critical domains. With ISO certification paving the way, AI’s integration into automotive safety designs becomes a tangible reality. Provizio envisions a future where trust in their 5D Perception® technology is so profound that it becomes as fundamental as a safety belt in a vehicle.

Provizio’s journey, fueled by innovation, collaboration, and a relentless drive for safety, is a beacon of hope for the automotive industry. The narrative they weave is one of technology embracing humanity, as they steer the wheel towards a future where safety and AI are seamlessly intertwined, setting a new standard for the roads that lie ahead.

Automotive Industries asked Barry, how Provizio’s collaboration with exida came about and what specific goals you aim to achieve through this pilot certification project for machine learning in safety-critical applications?

Lunn: We met with exida through our mutual collaboration with Future Mobility Campus Ireland, a purpose-built mobility test facility in Shannon, Ireland where we both have our headquarters. We were pushing the edge of radar-based perception using both Digital Signal Processing (DSP) and Artificial Intelligence (AI) and exida were pushing the boundaries of certification into the domain of machine learning so it was a very natural fit. Our goals were to certify our sensors and software to full automotive and industrial compliance but also to pilot the certification of our machine learning algorithms; this means our customers can benefit from the continual performance improvements over time as we train them on more and more data. So always improving perception while maintaining compliance with safety standards.

Automotive Industries: How does Provizio’s 5D Radar technology contribute to achieving ISO 26262 and IEC 61508 certifications, and what unique features or capabilities does it offer in terms of safety and perception in autonomous driving scenarios?

Lunn: Our multi-patented radar technology delivers unmatched resolution and range without the addition of costly additional radar channels, reduces interference and delivers robust performance in all weather conditions. This ensures improved raw radar performance which is the bedrock we build our DSP and AI chain on top of. So much higher repeatable and measurable performance allows us to achieve the relevant ISO standards. With this robust radar performance we are building superior perception performance, better classification performance for example, which allows us to accurately understand the scene in real time, predict future actions and prevent unsafe outcomes.

Automotive Industries: What challenges and considerations did Provizio encounter in integrating AI and Machine Learning 5D Perception® technology into the safety-critical path, and how is this pilot project addressing and validating those aspects?

Lunn: The adoption of AI into the 5D Perception® stack is more of a ‘practicality’ than an ‘ethos’. It would be possible to build a military-grade perception that could achieve the same outputs as our 5D Perception® engine, but if the sensors cost more than the car, who would use it beyond limited AV use? We really believe that this technology is a true path to safety so it has to be cost-effective enough to be widely adopted both for new builds, but also in the retrofit market. So AV perception at today’s ADAS price and form factor. With that decision made, the ‘practicalities’ of how, where and which algorithms should run, and how they could be trained had to be resolved. Ultimately this involved the collection of over 100,000 different training datasets taken over different environmental conditions. The pilot program validates these choices.

Automotive Industries: Could you shed light on the methods and techniques being utilized, such as FMEA and hardware FMEDA, in the detailed design review of the VizioPrime™ and VizioPlex Radar™ products, as part of the certification process?

Lunn: The technology is being assessed as a SEooC, with both an FMEA and FMEDA with DCA being utilised and is based around the underlying radar technology which can then be transitioned to various hardware builds and integrations.

Automotive Industries: How does achieving ISO certification for machine learning technology, through this collaboration, align with Provizio’s mission to improve road safety and move towards Vision Zero, and what benefits will it offer to both end-users and OEM partners?

Lunn: Provizio is always seeking to find the happy medium between safety and comfort and between performance and cost. Our commitment to safe outcomes is evidenced by this push into certification and we are doing this at the earliest possible point. We have seen the perception sensor hype raise billions of dollars and have zero impact on safety. That day is over and we are focused on delivering performance and that performance holding up to the scrutiny of safety certification. In a nutshell, this says we are setting the standard so nobody can say otherwise and OEMs can adopt the technology knowing the fundamentals are solid.

Automotive Industries: Looking into the future, how does this collaboration and the pursuit of ISO certification set the stage for Provizio’s role in shaping the integration of machine learning and AI into safety-critical designs and applications, particularly in the automotive industry?

Lunn: Although society trusts AI systems in many areas of modern society, such as flying, it is still reluctant to trust AI systems on our roads. However with the trust in the system that ISO certification will bring, we see the possibility of AI being accepted further. That means we have to deliver and qualify safer outcomes. It’s often stated that while many people are prepared to travel in a fully autonomous vehicle, they would not be prepared to let their loved ones travel in one. We want to reverse that thought, that it would be as ludicrous to get into a vehicle without 5D Perception® as it would one without a safety belt. ISO certification of an AI system is a major stepping stone along that path.