It could be said that today the Holy Grail of driving technology for the automotive industry lies in the possibility of finally developing an efficient autonomous driving system. It is a desire that all competitors, particularly Tesla Motorswish.
Elon Musk’s automotive company, despite selling itself as a highly avant-garde and innovative firm in this field, has constantly been criticized for the performance of its Autopilot system, which has triggered more than one fatal incident.
At the same time, for years there have been rumors and leaks of reports of active projects within Google and Apple where they would work at full speed to create fully autonomous cars in their driving.
But apparently there would be a foreign firm that would have thought of an even more revolutionary solution, with a system that could aspire to turn any vehicle into the means of transport of the future.
Seoul Robotics takes your car into the future
It turns out that a South Korean software company, Seoul Robotics, is using NVIDIA technology to turn ordinary cars into autonomous vehicles with an innovative system.
Because instead of placing its level 5 control tower, called the LV5 CTRL TWR, which is a mesh network of sensors and computers, on individual cars, the system is placed on infrastructure around a facility, such as buildings or utility poles. of light, to obtain a complete vision of the surrounding environment.
Seoul Robotics Vice President of Products and Solutions Jerone Floor thus outlines in the NVIDIA official blog the qualities of this kind of infrastructure:
“Instead of equipping vehicles with sensors, we are equipping the surrounding infrastructure with sensors. No matter how smart a vehicle is, if another car comes around a corner, for example, it won’t be able to see it.” The LV5 CTRL TWR towers give vehicles the latest bits of information gathered by taking a holistic view of the environment, so they are never driving ‘blind’.”
Obviously the cars would need adaptations of another type to be able to enable their autonomous driving system. But as we can see in the video, the system works at its most basic level.
Seoul Robotics’ LV5 CTRL TWR platform in its pilot version has been implemented in a BMW manufacturing plant in Munich and for now allows cars to move autonomously within the mapped terrain by compiling 3D data of the environment with the help of cameras. and lidar sensors.
The collected data is then analyzed using computer vision and artificial intelligence based on deep learning to determine efficient and safe routes for vehicles.
To make this possible, the NVIDIA CUDA libraries have been used as a base to produce GPU-accelerated applications mixed with cloud computing functions.
It just needs to work at a much more robust level.