Throughout history, humans have innovated to make mobility more efficient. Stemming from horses and horse-drawn carriages centuries ago, the auto industry emerged with the invention of the internal combustion engine (ICE) in 1860. Over the last few centuries, the ICE developed into a staple of our society – however, during the last few decades, car have been shifting to an electric vehicle (EV) model, starting with the hybrid revolution in 1997, and Tesla’s introduction of the roadster in 2008.
Over the years, the shift to EVs has transformed the auto industry – traditional manufacturing companies had to adapt to compete with technology companies specializing in battery-powered vehicles. In present day, the auto industry is experiencing a shift to autonomous vehicles (AV). Parallel to the EV revolution, manufacturing companies must adapt to compete with technology companies who are building self-driving solutions.
Currently, we are in the era of partial autonomy.
With minimal human supervision required, cars are able to change lanes, prevent forward collisions, and even navigate on highways. However, today, cars cannot fully operate under any road circumstance and environmental condition without a driver present. This is because traditional auto OEMs are using platforms based on legacy technology, such as the graphics-processing-unit (GPU).
Due to technological constraints, these solutions are not purpose-built for the monumental task of self-driving. Tesla, a technology company, is developing its own AI vision solution internally for their vehicles. Although they cannot equip their cars with full autonomy today, their autonomous capabilities are far beyond its competitors. To keep up with the rapidly innovating auto industry, traditional car companies must prioritize self-driving research and development – specifically, they must horizontally source a purpose-built solution to integrate into their cars to be competitive in the long run.
The biggest barrier to achieving vehicle autonomy is known as the visual perception problem. As you drive your car, your brain uses almost 100 billion neurons to flawlessly recognize and interpret your surroundings, while consuming a miniscule amount of power. To do the same on its own, an AV needs 75 Tera-Operations-Per-Second (TOPS) of compute for every watt of power consumption – an efficiency requirement not yet met by the industry.
To overcome the visual perception problem, an autonomous vehicles needs to be equipped with a solution that is purpose built for self-driving. This technology needs to be state-of-the-art, and must leverage key innovations in mathematics, ASIC architecture, and AI vision to have vast capabilities. Auto OEMs must integrate such a “best-of-breed” solution into their vehicles to enable vehicle autonomy. By doing this, they will be competitive and profitable in the long run as the auto industry adapts.
Picture: Cars Autonomously Driving/ Source: shutterstock
Ashwini is currently the Chief Business Officer at Recogni, the designers of a vision-oriented artificial intelligence platform for autonomous vehicles. He is a serial entrepreneur/company-builder with >20 years of experience and has built six startups with four exits via acquisition to various public companies.
Website link: www.recogni.com
Statements of the author and the interviewee do not necessarily represent the editors and the publisher opinion again.