Back to Media Center

The Complexities of Vehicle Autonomy 

The Complexities of Vehicle Autonomy 

By Sidhart Krishnamurthi, Product Management

Ever since the introduction of the ICE (Internal Combustion Engine) car in the early 1900s, the rate of innovation has rapidly increased, rending the impressive advancement of vehicles. In 1997, the industry released its first version of a “hybrid” car (Toyota Prius), and only a few years later, Tesla released the “Roadster,” the first commercially available electric vehicle (EV).

Along with the emergence of EV, vehicle autonomy has become a central focus of the automotive industry, evidenced by cars now featuring partially autonomous capabilities such as AEB and collision avoidance. Looking at the driver behind this — technology companies such as Tesla that are entering the automotive industry specialize in innovation with regards to vehicle autonomy. Unlike traditional companies in this space, these companies are vertically integrated — they prioritize tech R&D internally, allowing them to bring vehicles with advanced autonomous capabilities to the market at a much faster rate than incumbents. The industry recognizes this, and are scrambling to adapt in order to be competitive in the long run.

Today, we are in a state of partial autonomy. Specifically, the latest models of cars today are able to drive on their own on the highway, with the presence of a human driver ready to take over if needed. However, there is nothing on the market that can enable cars to be fully autonomous under any condition. Looking at the reason why — incumbent solutions in vehicles today are either general purpose processors or are based on computer vision. GPUs cannot be repurposed for self-driving — they are not built for such a task — and solutions based on computer vision do not have the compute efficiency to enable anything past partial autonomy. Clearly, the status quo is not suitable for high level AV perception, and the industry is stuck at partial autonomy as a result.

As elaborated above, technology companies entering the automotive sector are creating urgency within traditional companies who have been around for generations. As the industry goes electric and cars become more and more autonomous, companies who can innovate AV and battery technology the fastest will come out on top. Just look at the image below, which demonstrates Tesla’s rapidly increasing market presence given the societal transition to EVs.

Without Tesla, the U.S. EV market is faltering

The industry recognizes this and is prioritizing the development and integration of advanced AV technology to counter. Just look at recent acquisitions in this article or investments such as this one. Clearly, we are in the midst of a race to full vehicle autonomy.

When delving into the main reason why addressing this issue is so complex, one must look at the whole AV system. For the optimal level of accuracy, high-resolution, high-frame rate camera data must be processed in real-time under any condition. Today, there is nothing on the market that can do this, and the industry is trying to utilize a LiDAR based approach as an alternative. However, LiDARs are low-resolution, extremely expensive, and not robust under every condition (especially inclement weather), rendering them as rather unscalable for autonomy. The industry must push for a processing platform that is capable of handling 8MP camera data in order to develop self-driving cars.

We @ Recogni are developing such a platform. Building upon key innovations in ASIC architecture, math, and AI, we are building a solution from the ground up that can enable a car to drive from point A to point B without any human intervention under any condition. We are able to do this through our ability to process high-res, high-frame-rate camera data. The market opportunity is clear — there exists urgency in traditional car companies due to new entrant technology companies that can innovate at a much faster pace. Our product is a perfect fit for this opportunity, as we can enable car manufacturers to scale with the industry as it transitions to full autonomy, allowing them to be competitive in the long-run.


You May Also Like

3rd Party Media
3rd Party Media
Articles
Converting a network for the fastest inference chip on the market

Recogni Core-Tech Team *Conversion: Making a full-precision network compatible with our chip Introduction Deep Neural Networks (DNNs), in particular convolutional ones, are the go-to approach for visual perception tasks that…

Subscribe to Our Updates

Join our newsletter to stay up to date with Recogni!

Enabling ADAS to Full Autonomy