Recogni powered autonomous vehicles quickly and precisely process
Please introduce yourself and your startup Recogni to our readers!
My name is Ashwini Choudhary. I am a serial entrepreneur, with over 20 years of experience. Over the years, I have built six startups, with four of them exiting via acquisition. Currently, I am the Co-founder and Chief Business Officer at Recogni, a company based out of Silicon Valley. We @ Recogni are designing a vision-oriented AI platform purpose-built to enable autonomous vehicles to quickly and accurately survey their surroundings.
How did you get the idea of Recogni?
Back in 2013, I was riding my motorcycle on a sunny day through the backroads of the Bay Area. Unfortunately, as I was riding, I suffered a devastating motorcycle accident which posed the question: Is there any way vehicles can prevent such situations automatically?
Why did you decide to start with Recogni?
After multiple surgeries and many months of rehabilitation, I synced with a former colleague and a classmate of mine from college and asked them the same question above (Is there any way vehicles can prevent such situations automatically). As we did research, we figured out that there is no solution on the market today to enable fully autonomous vehicles (AV) to avert dire circumstances the dangerous collision I went through. This all culminated with the founding of Recogni in 2017.
What is the vision behind Recogni?
As you drive, your brain uses billions of neurons to generate 10000 Tera-Operations-Per-Second (TOPS) of compute. To mimic this and enable full vehicle autonomy, an AV would need to be equipped with a solution that has a ton of efficiency between computability and power consumption – specifically, it would need 75 TOPS of processing power for every watt of power consumed. This enormous constraint is known as the Visual Perception Problem
We @ Recogni are developing a purpose-built solution that will satisfy this immense performance requirement. By applying novel inventions in math and chip architecture, our product will have 1000 TOPS of processing power while consuming 10 watts, enabling AVs to see a traffic light 200 meters away and interpret visual cues in a few milliseconds.
In summary, our vision behind Recogni is to develop the only solution on the market that has the capability to solve the visual perception problem and enable fully autonomous vehicles.
How difficult was the start and which challenges you had to overcome?
Our biggest challenge stems from the fact that there are already industry incumbents backed by some of the biggest corporations in the world. These companies already have products integrated into vehicles that help with partial autonomy (Forward Collision Warning, Auto Emergency Braking etc…). However, we believe that we can overcome this, because our product will have unmatched technological capabilities, making it indispensable in the long run as the auto industry transitions to full autonomy.
Who is your target audience?
We are looking to sell our product to car OEMs (original equipment manufacturers). In order to adapt with the evolving auto market and be competitive in the future, these companies are looking for a solution that can provide them with the opportunity to scale from partial autonomy today to full autonomy in the future.
What is the USP of your startup?
The auto industry is facing an evolution to vehicle autonomy. With regards to the race to full autonomy, Tesla is the current industry leader, with their vertically integrated AI vision solution. Their platform has performance capabilities far ahead of solutions use dby traditional car companies.
Traditional car companies use a product developed by Mobileye, NVIDIA, or an accelerator to enable their vehicles with partial autonomy today. However, none of these solutions (and not even Tesla’s), has the ability to enable full autonomy – these platforms cannot meet the “75 TOPS per watt” requirement elaborated upon above. Given the evolution to vehicle autonomy, these companies are looking for a new innovation that can solve the future full autonomy problem, and inherently also the partial autonomy problem now. This will allow them to not only be competitive today, but also adapt with the market as it transitions.
We @ Recogni are developing that platform. As elaborated above, our solution will have 100 TOPS for every watt of power consumed, highlighting the fact that we can meet the processing requirement necessary to enable fully autonomous vehicles. Unlike current platforms used by car OEMs today, we can scale to full autonomy with our unmatched performance capabilities.
Long story short, our product is the only solution on the market today that can enable self-driving without human aid. Thus, we are the only product that can provide a path for traditional OEMs to adapt to the market as it transitions to full autonomy, allowing them to be profitable and competitive in the long run.
Can you describe your typical workday?
As the Chief Business Officer, my responsibilities vary across many disciplines. On any given day, you can find me doing a multitude of different cross-functional tasks. I work with the engineering side on a daily basis oversee the direction of the product. I am also heavily involved within marketing and business development serving as a main point of contact with potential customers.
Where do you see yourself and your startup Recogni in five years?
I want to have our solution adopted by every car in production at that time. Integrating our product should be second nature for car OEMs, given our product’s vast capabilities. Overall, I want our platform to move vehicle autonomy forward to make the roads a safer place to avoid dangerous collisions
What 3 tips would you give to founders?
Surround yourself with people smarter than you
Check your ego and be receptive to new ideas (keep an open mind)
Remember: “Hard work beats talent if talent doesn’t work.” No matter how smart you may be, you must outwork all your competitors.
More information you will find here
Thank you Ashwini Choudhary for the Interview
Statements of the author and the interviewee do not necessarily represent the editors and the publisher opinion again.