With projections of $91.83Bn by 2025, at a CAGR of 20.96%, ADAS (Advanced Driver Assistance Systems) features in the top 5 automotive super trends (Data source: MarketsandMarkets). Ensuring smart, secured, and convenient driving experience as well as providing multiple levels of independence, ADAS becomes another key driver of another super trend, which is autonomous driving.

While ADAS can be logically categorized as per functionality (Adaptive cruise control, Driver monitoring systems, Adaptive lighting system, parking assistance to name a few), or as per system components  (Camera/RADAR/Sensors etc.). However, from an engineering service provider’s perspective, it makes a lot of sense to look at ADAS more from an angle of how the data flows, processed, and how the actuation happens.

An ADAS from a service provider’s angle can be explained as a stepwise process in terms of components and data processing in the following manner:

Edge Devices / Sensors: ADAS takes input parameters from a large number of automotive sensors, including but not limited to camera, LiDAR, RADAR, gyroscopes, accelerometers, tire pressure, engine temperature, etc. This sensor data is used for a lot of analysis that help these driver assistance systems to decide on the further course of action. These actions could be passive (alerts) or active (actuations), but the continuous sensor data can certainly help in both regressive and predictive analytics, avoiding potential failures and accidents.

Connectivity & Pre-processing: The raw data from sensors is pre-processed for further analysis and provided to the server/ on-board hardware. One of the examples of this is image tuning and optimization for effective analysis. This pre-processed information/data is then sent to the processing unit over various automotive specific communication protocols such as CAN, Canbus, ETH100, MOST,Flexray, etc.

Data processing: Data processing which includes ADAS Kernels, Applets, ‘porting, and optimization’ of algorithms, development and validation of the software piece is carried out in this phase of the process. Various heterogeneous hardware platforms, like Hydra DSP, Multicore, multi OS SoCs are generally required for efficient processing of the data.

Sensor fusion: Data from various sources, though useful, its effectiveness grows multi-fold when viewed with reference to or in addition to data from other sensors. The data from multiple sensors, when fused together gives relevant insights into all internal and external factors necessary to generate meaningful actions.

Actuation: Finally, all the data and the conclusions drawn out of analysis of the same, needs to be put into practice in the live environment to actually reap the benefits of the technology for driver assistance.

This is primarily done in 2 ways:

• Warnings: Based on analysis, various alerts are generated suggesting the course of action based on the live scenario. This could be on driver’s drowsiness or parking assistance, safety alerts, Blind Spot Detection, Back Over Protection or Back Over Prevention, etc.
• Actuations: On the other hand, active safety or a more advanced form of ADAS would involve actuating the subsystems in case the initial warnings are ignored for some reason. The prime example of this is system actuating brakes by itself to slow down a vehicle to avoid collision.

Applications or Features of ADAS:

Advanced Driver Assistant Solutions find multiple ways to help the drivers in hassle-free driving and parking. It also ensures safety and security of drivers, co-passengers, as well as other drivers, pedestrians, and property. This safety is ensured with the help of vision based algorithms such as pedestrian detection, object detection, human Vs. Non-human detection, read signs detection, and reading as well as other useful traffic insights.

It can also detect and report traffic violations through other algorithms such as, a vehicle taking a wrong turn or vehicle needing assistance or detecting other vehicles, which are wrongly parked, or other such scenarios.

Apart from video analytics, voice integration, integrated cockpit powered by advancements in displays are also the trends that we will see in the near future.

On the safety and security side, not only collision avoidance systems, but also features like optical self-diagnostics, blind spot detection, night vision, lane departure and anomalous driver behavior alerts are some features that brings truly tension-free experience to the drivers and the passengers!

Choosing the right platform:

Choosing the right hardware for your safety-critical industry applications is very important. Various considerations are heavy multimedia processing, high edge memory, and computing power, among some others. Your PC based legacy algorithms cannot be ported as it is on the automotive platforms and needs a lot of customizations and code conversions before porting for optimized performance.

It needs expertise across sensor integration to communication protocols to ‘multimedia and displays’ to cloud, as well as edge based analytics to help the OEMs/ Tier-1s implement a state of the art ADAS system.

This is where einfochips’ expertise can add great value to automotive companies looking for advanced ADAS systems development. eInfochips has proven expertise in PC based algorithm porting on automotive chipsets, implementing video analytics features such as Object detection, pedestrian detection, camera calibration, blind spot detection, human vs non-human, vehicle counting and classification, and beyond.

Get in touch with our automotive expert today!