ADAS — Advance Driver Assistance System

Introduction

Let me introduce you to the driver of modern cars Advance-Driver Assistance System (ADAS). It is basically an electronic system that assists the driver in driving and parking the vehicle. ADAS is a self-learning interface that is used to increase road safety and minimize accidents on roads. It works on several autonomous levels depending on the technology installed in a car. ADAS with the help of sensors and camera detects the obstacles such as other cars, humans, animals, or any such thing nearby the vehicle and responds to them accordingly.

History

Back in the 70s in America ADAS was first available with the implementation of ABS (Anti-Lock Braking System). Further than more features were introduced including electronic stability control, blind spot information systems, lane departure warnings, traction control, and adaptive cruise control.

Levels

ADAS has featured over 6 levels starting from Level 0 to Level 5 depending on the autonomous features installed in a car. But you might be a little surprised that we have achieved level 2 till now. Although, Mercedes-Benz has recently received approval from the U.S. government for its level 3 ADAS this year. However, that technology isn’t expected to be accessible until 2024, as an option in the Mercedes Drive Pilot.

The several levels of ADAS are as follows:

1. Level 0: No control over the car

In this level of ADAS, the system has no control over the vehicle. The working of the vehicle solely depends on the driver. It can only offer the basic necessary info to the driver; the driver has to interpret the information and has to take action accordingly. This level of ADAS includes features like parking sensors, surround-view, traffic sign recognition, lane departure warning, night vision, blind spot information system, rear-cross traffic alert, and forward-collision warning.

2. Level 1: Single Functionality

This level is the beginning of automation in cars. It is a single automation function that features the steering or cruise control (only acceleration). While any one of the tasks is being performed by ADAS, the driver needs to take care of the other controls of the car.

3. Level 2: Multiple Functionality

The most common and featured level of ADAS today is level 2. It features the multi-tasking in car. The car can perform acceleration/deacceleration and steering simultaneously.

4. Level 3: Partial Automation

A level ahead from just acceleration and deacceleration it brings autonomous features like adaptive cruise control and lane centering. The system is capable of performing such tasks without human interventions in specific conditions but the driver should be ready to take control over the vehicle anytime.

5. Level 4: A closer step to full automation

It is far better than level 3 automation. It is categorized as the higher order automation technology which doesn’t require human intervention at all in specific conditions only. It can perform all aspects of driving without the presence of a human driver in specific weather and driving conditions. While it can’t be categorized as a fully automated system as it faces problems in urban conditions.

6. Level 5: Fully Automation

The last level of ADAS provides the user with driverless vehicles which can operate in any condition, rather it is a traffic jam, bad weather, or any other such condition. But, till now such technology hasn’t been invented.

Features associated with ADAS

1. Adaptive Cruise Control (ACC)
2. Anti — Lock Braking System (ABS)
3. Automated Parking Feature
4. Lane Keeping Assist (LKA) / Lane Departure Warning (LDW)
5. Automatic Emergency Braking (AEB)
6. Blind Spot Detection (BSD)
7. Cross-Traffic Alert (CTA)
8. Traffic Sign Recognition (TSR)
9. Driver Monitoring System (DMS)
10. Forward Collision Warning (FCW)
11. High Beam Assist (HBA)
12. Pedestrian Detection
13. Traffic Jam Assist
14. Adaptive Headlights
15. Collision Avoidance System
16. Gesture Control
17. Emergency Assist

Role of Artificial Intelligence (AI), Machine Learning (ML), and Deep Neural Networks (DNNs) in ADAS

Photo by Google DeepMind on Unsplash

1. Sensing: Deep Neural Networks, a subset of ML, excel in image and pattern recognition, enabling accurate object detection, lane tracking, traffic sign recognition, and pedestrian detection.
2. Sensor Fusion: AI integrates data from multiple sensors to create a comprehensive understanding of the vehicle’s surroundings. ML models fused data from different sources to provide a more accurate and reliable perception of the environment.
3. Decision Making and Control: ML algorithms analyze sensor data in real-time to make informed decisions about vehicle control. AI-driven controllers manage tasks like adaptive cruise control, lane keeping, and emergency braking.
4. Mapping And Localization: AI-driven mapping systems provide accurate maps that help vehicles navigate. ML is used for precise vehicle localization on these maps, allowing the vehicle to know its position and plan its path.
5. Predictive Analytics: ML models analyze historical and real-time data to predict traffic patterns, road conditions, and potential hazards. This enables the vehicle to anticipate and react to situations more effectively.
6. Driver monitoring: AI-powered systems monitor driver behavior and attention to facial recognition and eye tracking. ML assesses the driver’s state for potential drowsiness, destruction, or impairment.

7. Anomaly Detection: ML algorithms identify deviations from normal behavior, alerting the system to potential failures or anomalies in sensors or components.

8. Updates and Improvements: ML algorithms enable over-the-air updates, allowing vehicles to receive software improvements and adapt to new driving scenarios without requiring physical updates.

9. Training and Simulation: ML models are trained on vast amounts of real-world and simulated data to improve their decision-making capabilities. Simulations allow the testing of countless scenarios, enhancing system reliability.

10. Constant Learning: AI and ML systems in ADAS can learn from new data and experiences, adapting to new situations and improving over time.

11. Complex Pattern Recognition: DNNs excel in identifying intricate patterns and correlations within data, making them valuable for understanding complex traffic scenarios and making split-second decisions.

12. Reducing False Positives: ML models help reduce false-positive alerts by discerning between real threats and harmless events, improving the system’s trustworthiness.

“ADAS is also related to our safety, want to know how, then stay tuned for more!!”