You’d be mistaken to think, ‘Why all of a sudden, autonomous driving?’ The history of autonomous driving (AD) goes back nearly a century at the longest, or about 40 years if you look closely. Early challenges for autonomous driving began in the United States, Germany, and Italy. In Korea, the first AD car developed with domestic technology was born in the early 1990s. Coincidentally, all of them are automobile powerhouses that have dominated in the past or dominating the global market now. It also means that their desire for the “safest road and transportation system” was that great.
In the United States, the Defense Advanced Research Projects Agency (DARPA) intervened and accelerated AD research, and in Germany in the late 1980s and early 1990s, the concept of AD in a modern sense appeared. In the DARPA Challenge (crossing 240km of the Mojave Desert by AD), which first started in the United States in 2004, not a single car was successful in the first year, but the following year, the Stanford/Volkswagen Stanley team completed the race in 6 hours and 54 minutes. They could finally lay a foundation for realizing the dream. This later led to the starting point of Google Waymo.
The actual AD craze began around 2010 with the development of artificial intelligence (AI) technology. Now, technology-based companies such as Waymo, Tesla, Baidu, and Intel/Mobileye, as well as many companies such as Hyundai, GM, Mercedes, and Toyota, are challenging to realize L4-level of AD beyond state-of-the-art ADAS (advanced driver assistance systems).
The future will be bigger than the present… The key is the core technology.
The most important prerequisite to AD tech is perception. It must accurately identify all objects surrounding the car, including obstacles, other cars, human beings, and stationary objects around it. You must be able to precisely calculate the distance to the target and at the same time accurately check your location to calculate the approach time as a result. It also needs to detect whether the target is stationary or moving, and it needs to detect the traffic system such as lanes and signal lights on the road. In a word, ‘perception’ is the most important basic and key prerequisite for the implementation of advanced ADAS and AD.
Once you have finally completed the ‘perceive’ phase, you must make a decision (Planning & Decision) on what action to take based on what you have recognized. Once you have perceived an obstacle, you must decide whether to stay in your lane or change lanes. There are situations that require judgment and decision almost every moment on the road, such as when to make a turn, whether to overtake or follow the car in front of you. Once the decision is made, control follows. Here, acceleration and deceleration, steering operation, and gear shifting are executed according to the route saved in advance by the driver, and actual driving continues.
To carry out all these key steps without problems, proper technical support should follow. Advanced technologies such as LiDAR, high-performance sensors, radar, and cameras are essential to perceive the surrounding environment. Lidar recognizes a 360-degree area around it, and a high-performance camera distinguishes traffic signals, nearby cars, and pedestrians. Ultrasound and radar recognize approaching vehicles or front and rear to detect the approach of risk factors in advance. All the important information is linked to the wheel encoders that are mounted on wheels to determine the amount of distance traveled or inertial measurement units (IMUs) that calculate the amount of movement by measuring acceleration to induce optimal driving. And now, artificial intelligence (AI) technology is added to this. By tracking an object with AI technology, recognizing an image, and predicting the surroundings in advance based on it, it takes a step toward fully AD.
Based on the L2 level of advanced ADAS, AD is already steadily stepping into the realization stage. This year in 2023, the market for AD vehicles at the level of L3 and L4 is expected to reach $6 billion based on privately owned vehicles. If we move on to the current trend, this will grow to a whopping $77 billion dollars in 2035. It’s still too early to be surprised. The scale of related businesses such as cameras, AI software, lidar, radar, and maps is likely to grow even larger than this. There is also an expectation that the scale of these related businesses can grow to $98 billion in 2050.
STRADVISION’s role in the ADAS and AD market
STRADVISION, an AI start-up for AD solutions, is the protagonist who developed ‘AI deep learning-based object detection software (SVNet) technology’, a key technology for the popularization of ADAS and commercialization of AD. In the field of advanced ADAS and L2 or higher AD, almost all automakers use different systems, which is why STRADVISION’s fast and lightweight deep learning software shines even more. ‘Ultra-light and high-efficiency’ that implements deep learning-based object detection with minimal computation and power consumption, and ‘flexibility’ that realizes excellent object detection function based on AI even in expensive chipsets as well as low-priced automotive chipsets are the core SVNet. Even if a new SoC comes out, deep neural network-based object detection can be implemented within an embedded platform in a short period such as less than six months.
The fact that it can be implemented in any SoC, and that high-performance camera recognition software can be applied without incurring a large cost is not only a key competitive advantage that differentiates it from existing competitors such as Mobileye, but also a critical solution point that appeals to automakers and global tier parts makers.
As of 2022, the number of vehicles equipped with STRADVISION’s SVNet is 334,000 units, which is expected to grow to over 42M units of volume by 2032. Considering the outstanding strength of SVNet, which secures the highest level of safety by using single and multi-cameras to accurately perceive various objects such as cars, pedestrians, cycles, and traffic signals, this prospect is highly likely to become a reality. If AD technology is the key to unlocking the future, at least until around 2030, the automotive camera and perception technology market will grow at an incredible pace.
Autonomous driving is the core of future roads, and object detection technology is the main player.
STRADVISION is also one of the key players accelerating the ‘realization of the future’ when it comes to automotive usage. With the help of ADAS, the benefits that can be gained by cars taking over the role of humans on the road are greater than expected. First of all, the number of traffic accidents that are directly caused by drivers will be significantly reduced, which is 70% (as of 2019), and traffic congestion will be minimized by finding and operating the optimal route. Ultimately, the whole process will make a great contribution to the environment as well. There is also a reason why many companies that are in a hurry to introduce AD technology are interested in robotaxi services. A whopping 95% of the total usage time of a car is parked. It can be a waste. The ‘proactive, future-oriented and most realistic approach’ that can solve this problem and operate it most efficiently is the robotaxi.
To learn more about STRADVISION and their SVNet software technology, please visit www.stradvision.com.