What is ADAS technology in vehicles?

With the development of society, driverless technology continues to mature, and advanced driver assistance system (ADAS) functions have been proven to reduce accidents and save lives. According to the Insurance Institute for Highway Safety in Consumer Reports, there were 50 percent fewer front and rear crashes in cars equipped with forward-collision warning and automatic emergency braking systems compared to cars without them in 2017. Unfortunately, most accidents happen to vehicle owners who do not have even the most basic ADAS applications installed.

As ADAS continues to evolve toward Level 4 and Level 5 autonomous vehicles as defined by the Society of Automotive Engineers, we have the opportunity to have an even greater impact on the road by creating autonomous vehicle technology that can be used in a wider range of vehicles.

Although it is economically impractical to equip all cars with full ADAS technology, the goal should still be to equip as many cars as possible with driver assistance features. This means more vehicles on the road need to be able to efficiently sense, process and respond to real-time data.

The need for smart and diverse sensing

Traditionally, image data collected for ADAS operations has been analyzed by feature-based computer vision algorithms. Computer vision has served the industry well over the past decade, but as ADAS operations become more advanced, designers need additional tools to handle and adapt to the situations drivers and their vehicles face on the road.

Keeping ADAS operational under different road conditions is a challenge. When encountering unexpected situations such as bad weather or poor road conditions, vehicles need to adapt in real time. These situations are difficult to handle with traditional models, but by developing a dynamic system that helps the car sense, understand, and react quickly to the world around it, the car itself can become an effective co-pilot for the driver. Such a system requires data and the ability to process it in real time using a combination of computer vision and efficient deep learning neural networks.

ADAS solutions need to extract data from different sensor sets and convert the data into behavioral intelligence for the vehicle. At a minimum, these sensors will need to be equipped with different types of cameras and associated optical, radar and ultrasonic technologies; in more complex cases, lidar and thermal night vision devices will also be required. Additionally, the system can localize vehicles by comparing features extracted from sensor data with high-definition map data. Understanding and analyzing this multimodal sensor data must occur in real time (new data arrives 60 times per second) without the need for a data center server in the back seat of a car.

Learn how to perfect autonomous parking technology with the Jacinto™ processor.

The solution must be road-ready

Drivers must receive multiple pieces of information simultaneously and make safe driving decisions quickly, and so must all ADAS applications, regardless of autonomous driving level. The importance of a high-performance system-on-chip (SoC) is that it enables parallel processing without requiring significant budget cuts in power, temperature, component and integration costs. SoC solutions can scale from simpler cases (fewer sensors, lower resolution) to the most complex cases without compromising basic ADAS functionality or requiring downgrading the system level.

Adapting to the application performance of various types of vehicles is only one of the requirements. The development of the system must be cost-effective in order to achieve widespread and effective utilization. The complexity of in-vehicle software is growing exponentially (today it is 150 million lines of code), which is causing development and maintenance costs to skyrocket. As systems become more aware of road conditions, their functional safety requirements will continue to change and evolve and must meet stringent vehicle quality and reliability targets. It is these strict requirements and realities that support and drive the development of the automotive electronics market.

The right SoC can address all these needs. It can appropriately balance memory, input/output, and processing cores based on a range of application requirements to achieve the system's BOM goals. The right SoC can also accommodate open software development methods, making it possible to use generated code multiple times, saving effort in development and testing. SoCs can also be built with functional safety in mind from the start, and have the necessary reliability and product longevity to allow automotive production lines to last in the market for many years. As long as we do it right, equipping more cars with powerful ADAS functions (as shown in Figure 1) is just around the corner.

How TI is helping democratize ADAS technology

TI leveraged decades of automotive and functional safety expertise to design our JacintoTM 7 processor platform to solve sensing, parallel operations and system-level challenges.

We focus on aspects that have an important impact on the entire system: combining excellent perception capabilities with multi-directional monitoring of the car's surroundings and adopting a car-centric design approach to optimize power and system costs.

The new JacintoTM 7 processor family, including the TDA4VM and DRA829V, integrate key functional safety features on the chip to enable both safety-critical and non-safety-critical functions in a single device; they also improve data by combining high-speed and automotive interfaces manage. The JacintoTM 7 processor brings real performance to automotive ADAS and gateway systems and helps reduce system costs, thereby democratizing and popularizing ADAS technology. The above is some knowledge about ADAS driving technology. I hope it will be helpful to practitioners learning driverless driving.