Q&A with Recogni CEO and co-founder RK Anand
Part of the Toyota AI Ventures portfolio since 2019, Recogni is powering autonomous driving technology. Its visual perception platform uses a high-powered chip with artificial intelligence (AI) to efficiently enable vehicles to perceive their environment. Recogni recently raised a $48.9 Series B financing round, and we sat down with co-founder and CEO RK Anand to hear more about Recogni’s technology and what the Recogni team is aiming to do next.
Thanks for joining us, RK. Tell us about your journey to co-founding Recogni.
Thank you to Toyota AI Ventures for giving me this opportunity to tell the story of Recogni. Throughout my career, I have been fortunate to be a part of great companies like Sun Microsystems and Juniper Networks, building innovative products and making breakthroughs in advancing state-of-the-art technologies. Sun was where I got my basic understanding of robust engineering practices — learning from great mentors and seeing exceptional and motivational managers. At Juniper, where I had the good fortune to be at a startup from the beginning, I got to see things go from idea to realization with rapid growth, transforming markets and industries. Most importantly, once in leadership roles, I got to give back to the next generation of talent and see them blossom and thrive.
Recogni, in my humble opinion, has the opportunity to be another one of those category leaders — through sheer innovation, architectural insight, and creativity. A great team has been coming together that is very capable across many disciplines: silicon design, AI, embedded and platform software, system design, and a whole lot more. We are building the fundamental AI edge compute necessary for exceptional visual perception, which is the prerequisite and foundational need for safe autonomous vehicles.
How does Recogni’s visual perception platform work?
At Recogni, we believe that perfect perception is the essential and critical element of any autonomous system. Vision is how we as humans navigate the world and drive vehicles. It has to be the primary sensing method for computationally driven vehicles — primarily because of the richness of information available. Secondary sensing methods like radar and LiDAR offer diversity in information — but neither has the density or quality of information of vision needed.
As the resolution of cameras increases from 0.5 megapixels (MP) to 8 MP and more, and frame rates increase — the compute needs increase non-linearly. The higher the resolution and frame rate, the better the quality of what is perceived. Recogni’s purpose-built perception is designed to tackle this massive increase in compute with real-time results — so that autonomous vehicle manufacturers can overcome the barriers they face today in making their vehicles truly capable.
Across the autonomous driving landscape, there are many different approaches regarding perception systems and sensor platforms. What is unique about Recogni’s approach?
Our view is that perfect visual perception is the required foundational element for any autonomous system. It is on this foundation that the superstructure of autonomous software has to be built. Recogni has the advantage of building upon the incredible work of other companies and researchers who came before it and advanced the science and capabilities until today.
A lot of the approaches were built on legacy and general-purpose computational platforms. Recogni took a clean-sheet approach to the compute challenge for visual perception, and with our architectural innovations, we are building a computational solution that will help automotive companies get to the true realization of autonomy.
The Recogni platform and its chip are built to have high processing power, while simultaneously consuming very little power. Why is this important, and can you share — on a general level — how the team accomplished this?
As inexpensive camera sensors increase in resolution and frame rates — the compute burden of processing this data is quite significant. If technology companies do not provide this exceptionally high AI inference compute capability at very low-power consumption and an inexpensive cost, then we have a challenge as an industry. Ubiquitous safety and autonomous vehicles will have significant societal benefits — the vehicles cannot have compromised range, else the utility will be diminished.
The very capable Recogni engineering team is versatile across multiple dimensions — from chip architecture to design, from hardware modeling to system software and all aspects of AI training and inferences. We had to bring all these skills together in a seamless way to get us to this point and further.
Under what kind of road and weather conditions does Recogni enable an autonomous vehicle to see, and how quickly can the vehicle react?
We expect autonomous vehicles to operate in all road conditions, urban to highway settings, in every possible weather scenario — be it sunny days or thunderstorms or low-visibility fog. We drive this way — why expect any less?
For all of these scenarios, near-perfect, real-time, and uncompromised visual comprehension of the environment in and around the vehicle is extremely important. With Recogni’s technology, we strongly believe that auto manufacturers will get a step-function in perception capability over what they have today, getting them much closer to the realization of true autonomy.
Congratulations on completing your Series B financing round. What’s next for Recogni?
We are fortunate to have our new Series B investors — WRVI, Mayfield Fund, Continental Automotive, and Robert Bosch Ventures — invest in us. Our Series A investors, including Toyota AI Ventures, all came back strongly in this round too. All of this validates our thesis and product strategy. We expect the Series B investment to enable us to get the product to market this year, support customers, scale our go-to-market teams, and bolster engineering.
Do you think the pandemic has impacted the way the general public sees autonomous vehicle technology? Will it cause autonomous technology development to accelerate?
The pandemic in 2020 and continuing into this year has fundamentally and irreversibly changed how a lot of us function, and will function going forward. There are areas of our lives where we will not go back to how we behaved during pre-pandemic times.
We expect over the next few years an acceleration of robo-delivery and robo-taxi systems. We already see major automotive companies almost overnight totally embracing EV as the way they will go forward. These shifts don’t happen often in our lives, but when they do, the speed of the changes is rapid — as both human ingenuity and capital get focused on the task and incredible innovations happen quickly.
Recogni is poised to take advantage of this acceleration towards autonomous vehicles.
Recogni is partly based in San Jose, California. Much has been said about tech companies moving from the Bay Area. What are your thoughts on this trend, and what are the plans for Recogni?
Recogni has the advantage of having our teams in both San Jose, California, and Munich, Germany. Let us not forget the incredible and experienced talent pool that live and thrive in the greater Bay Area — and the fantastic companies that originated here.
The Bay Area has gone through many cycles and has always come back stronger, with new creativity and vigor.
In my personal opinion, this “moving from the Bay Area” is just another fad that has gotten press, focusing on people with immense wealth and social-networking reach. The majority of the talent still resides here, innovates and works every day, and moves the needle of many of the companies competitively.
Where do you see autonomous vehicles — and Recogni — in five years or so?
The journey to fully autonomous vehicles is a continuum, and we expect Recogni to be an essential piece of the solution for many of the automotive companies. Technology has a unique and elegant ability to creep up on us — and all of a sudden, we’re wondering how magically things came to be. But as with everything great, a lot of hard work, failures, and breakthroughs go into it.
Visit the Toyota AI Ventures site to learn more about Recogni and the rest of the TAIV portfolio.
