The Vision Is Getting Clearer… But Vehicles Still Have Trouble “Seeing”

February 15, 2022

It’s been exciting to see sophisticated driver assist (ADAS) more broadly coming to market from the leading car manufacturers, with “hands free” automated driving features around the corner. The same is true regarding the impressive progress from the companies providing autonomous trucks, delivery robots, and robotaxis. The annual Consumer Electronics Show (CES) earlier this year again put a spotlight on these capabilities and the technologies that power them. We’re excited to mention that at the CES Innovation Awards, Algolux Eos perception software was recognised and awarded.

But despite the progress to date, the effectiveness of these ADAS and automated driving capabilities still have a way to go and this problem needs to be addressed ASAP. This provocative video by tech analyst Patrick Moorehead show Tesla’s automatic emergency braking (AEB) system failing to brake for a child test dummy, such as those used for testing vehicle safety.

Computer Vision Fail CES 2022 Tesla - Here is what Algolux has to say

This is further reinforced by some recent studies. The Wall Street Journal just reported that car fatalities in the US rose by the fastest rate in half a century, over 30,000 car-crash deaths in the first nine months of 2021, mainly due to driver actions. The Insurance Institute for Highway Safety (IIHS) show significant issues with AEB across many vehicles using different perception approaches, especially in dark conditions. So while driver assistance systems could clearly mitigate these terrible statistics, these systems are currently not robust… they don’t “see” that well in general, and hardly at all at night and in poor weather conditions. 

Autonomous vehicles and especially autonomous trucks can afford many more sensors, including numerous cameras and Lidar, to provide a better view of what’s in front and around the vehicle. But there are still limitations in difficult weather and lighting conditions. Also, a fully loaded tractor-trailer weighing 80,000 lbs can take hundreds of meters to avoid hitting a stopped vehicle or dangerous debris in the road and even state of the art Lidar pragmatically maxes out at about 200m.

The move to automated driving and autonomous vehicles will optimize our travel, saving time and lives. But while the lack of robustness to low light, poor weather, and distance is a well-recognized issue by the industry, breakthroughs are needed to realize this vision of travel and transport. Algolux is fundamentally focused on solving this problem with our leading AI and computer vision technology and it is great to see the recognition at CES of what our team has accomplished so far. This year promises to be a very exciting one for us at Algolux, so keep watching as we will be sharing more on new technologies and solutions. Drive safely!