Autonomous driving is the next big thing slated to disrupt the automotive industry after electric and alternative fuel cell cars. This technology aims at reducing human intervention in driving, hence reducing the fatalities and accidents caused by human error, one of the major reasons for road accidents. Autonomous cars and their capabilities were not built overnight, but all the technologies that facilitate autonomous driving were under development from as early as the late 20th century. Technologies like cruise control, park assist, etc., were initially found only in cars that would cost a fortune. These technologies later were adopted into general passenger cars to make driving safer and more enjoyable.
Understanding autonomous cars and levels of autonomy
Autonomous cars are often misunderstood to be synonymous with driverless cars. But in reality, they are not. There are various levels at which cars are classified as autonomous. This classification is based on the amount of human interaction involved while driving.
- Level 0(L0) - No automation and the driver is in control of all aspects of driving and controlling the vehicle.
- Level 1(L1) - Driver assist systems like adaptive cruise control and emergency braking that are employed to aid the driver in reducing the effort of mundane highway driving.
- Level 2(L2) - Partial assist systems in which a car can control steering, braking, and emergency acceleration.
- Level 3(L3) - Along with the assist systems seen in level 2, complete surround monitoring and driver monitoring solutions (for drowsiness and fatigue).
- Level 4(L4) - Car is fully autonomous but requires presence of a driver to intervene in case of emergency.
- Level 5(L5) - Fully autonomous car that does not require a driver to be present. Cars can predict their path and reach their destinations safely.
Figure: 1- SAE levels of autonomy
Enabling sensing technologies behind Autonomous cars
It is safe to say that advances in sensor technology and improvements in the computational power of processors have fuelled the growth of autonomous driving. Some of the key sensor technologies that are driving adoption of autonomous cars are:
- Lidar- Navigating a busy street without being involved in any altercations is one of the major hurdles of the autonomous car. Lidar, even though invented in the 60s, is now finding newer applications in cars for object detection and mapping the surroundings.
- Radar- Initially developed and used in warfare in the early 20th century, radar is finding its way into new applications every day. Similar to lidar, radar is also used for object detection and collision avoidance. To understand more about radar and its applications in autonomous driving, click here.
- Camera/computer vision- Cars are equipped with a camera not only for park assist but also for surround view, the same cameras are used by systems to monitor the surroundings of the car. The system uses machine learning to identify various objects like traffic signs, pedestrians and other traffic. To understand how machine learning helps in autonomous driving, click here.
Seven technologies that are accelerating adoption of autonomous driving
There are numerous new and existing technologies are that being used in passenger cars. However, only a few are really making any impact on autonomous vehicles. We understood the sensing technologies that are in use, now let’s have a look at the accelerators and concepts that are using them to drive innovation in autonomous cars.
1. SLAM (Simultaneous localization and mapping)
This is a method of identifying the precise location of an object in an environment, be it indoor or outdoor, with varied weather conditions. This technique uses sensors to identify stationary objects and maps the surroundings accordingly. Autonomous cars when on the road cannot completely rely on GPS data to identify its position, since the accuracy of GPS systems is really not suited for driving in traffic, autonomous cars use both GPS and SLAM technology to identify precise locations and manoeuvre in traffic.
2. Machine Learning
If autonomous driving is a new phenomenon in the auto industry, machine learning is a new phenomenon in utilizing statistical tools and available data to predict results with greater probability. Machine learning has various applications in the products that help autonomous driving (some of them listed above).To understand how machine learning is changing autonomous driving and the automobile industry, Click here.
3. V2X communication
When vehicles reach full autonomy (SAE-L5) communication becomes one of the major factors for the easy flow of traffic. V2X involves vehicle to vehicle and vehicle to infrastructure communication, wherein vehicles are in contact with infrastructure like traffic signs and other vehicles in the path. This technology is enabled by IOT. By streamlining the flow of traffic, vehicles will not be wasting valuable time being idle in traffic. With V2V communication vehicles interact with other vehicles and can plan the route to reach the destination in the least amount of time.
4. IOT
With existing IOT technology connecting cars is easier. All the information of the vehicle and its users can be stored in a cloud for further processing, making the use of autonomous cars more efficient.
5. Telematics
Integration of wireless technology along with location and vehicle electronics can be termed telematics. Currently telematics applications are widely used in trucking and fleet management organizations to keep a tab on the user/driver. Hardware required for a telematic systems usually comes in new cars with free services like remote locking, road side assistance, and other features for a limited period and can be extended based on subscriptions.
Services like vehicle collision information is at the heart of vehicle telematics. When a vehicle’s airbag is deployed, the sensor activation is picked up by the telematics system, and with this live information location details for emergency services are shared.
6. Brain to vehicle technology
This is something radical with only a few manufacturers already testing their technology. A user’s brain waves are monitored by the vehicle to understand, predict, and eventually anticipate a driver’s behaviour. Since this technology involves wearing a cap-like instrument with electrodes attached, how comfortable users will be with that it is yet to be determined.
Edge Computing
Data collected from various sensor apparatus in applications like V2X was moved to cloud for further processing, as the amount of data increased day by day, latency and other issues started to hinder the desired performance. To address this issue one can, employ edge computing techniques to achieve desired results. Edge computing refers to performing limited and possible processing at the source rather than relying on cloud. Understand edge computing in brief here. Having edge computing technology in autonomous cars reduces its dependency on cloud and other high investment infrastructure and helps in improving their performance.
To summarize
Autonomous cars are not a thing of the future. They are here, even though various nations are yet to come up with the necessary legislative norms for testing the cars and actual implementation. Autonomous cars are advancing various innovations and also are poised to make the roads of the future a lot safer for both travellers and pedestrians.
Further reading
Interested in Radar Automotive application?
Here is our radar series of blogs explaining Automotive Radar
- Understanding Radar for automotive (ADAS) solutions
- Why are automotive radar systems shifting to 77GHz
- Modulation techniques for automotive radar application and why FMCW is winning the race
- FMCW CHIRP configurations for SRR, MRR and LRR
- Brain to vehicle technology as demonstrated by Nissan here.
Interested in Radar service and SDK?
