Self Driving Vehicles – How Do They Work? Explained

  • Twitter
  • Facebook
  • Google+
  • Pinterest

Self-driving cars are developing at a rapid pace and it seems like soon they will become the future of transportation. Major technology companies like Lyft and Waymo, as well as automaker companies like Toyota, have invested billions of dollars to manufacture and improve the technology of self-driving cars.

There is no doubt that the market for them could one day be worth trillions of dollars. Autonomous buses and shuttles are currently being deployed in cities and airports. Driverless trucks are already delivering products long distances and some manufacturers are even working on autonomous flying taxis.

You might think that why would we need self driving vehicles? There are many good reasons for this revolution. The self driving cars will greatly reduce the price of transportation for consumers since there is no human in the loop. Also, by using the autonomous fleet of shared electric cars we will be needing only 10% of cars on the road compared to the present.

This will also greatly reduce CO2 emission which is, of course, beneficial for our environment. Perhaps the most important reason out of all is that it would create a safer environment for everyone. Data from the National Highway Traffic Safety Administration shows that more than 90% of car accidents are caused by human error. This means that this technology can help save many lives in the future.

But how does this self driving technology actually works? Well to make it simple this tech works on main components which we have discussed below in the right order.

Computer Vision:

The computer vision is the step in the chain, it is how cameras see the road compared to humans. We, humans, handle the vision problem by handling a car’s steering wheel with just our two eyes and a brain.

The Self driving cars utilizes camera images to determine the lane lines and tracks down other vehicles on the street. They are equipped with multiple high-tech cameras some of them feature up to 8 cameras providing 360 degree vision.These cameras are capable of lane finding, obstacle detection, road curvature estimation, traffic light detection etc. But what if a pedestrian is about to cross the road, the car will first determine and map the distance of how far it is.

Now it will classify whether it is some kind of object or an actual person. The deep learning is the solution here it has come out as the most reliable approach to work with camera images and videos in this type of scenario. It trains a neural network by making it learn how a stop sign looks like by feeding thousands of stop sign images which will help in gradually learning its abstract representation.

Sensor Fusion:

The camera is not the only type of sensor that car features. The sensor fusion collects the data from all the sensors and builds up a complete understanding of the vehicles surrounding. Some sensors can work better for a specific task like in calculating distance and velocity while some are better in detection.

Like a Radar will help in determining how far the object is and how fast is it moving but this sensor can not classify the type of object and that is where lidar comes in, it shoots a pulsed laser beam which develops a 3D point cloud. So lidar is basically the essential component between camera and radar which helps in completion of the tracking and detection function.

Image: Self Driving Car/ medium.com
Image by the Medium

Another important sensor that you must know about is the Ultrasonic sensor that is designed to sensor small distance to help in lateral movements like parking. So by combining all the sensor data, we get a complete state estimate of the environment.

Localization:

The next step in the chain is localization which kind of works like a GPS that we have in our phones but is more advanced. Basically, GPS is only able to track within two meters accurately and if a car were wrong by that much it could result in fatal accidents.

The self driving cars use very advanced algorithms which allows seeing the point cloud that the map has and measures the vehicle distance to specific landmarks around it.

Path Planning:

Moving on to the next step, autonomous vehicles use path planning that charts through the map to reach the destination. It uses AI to predict what the other vehicles around it will do and then it takes the decision accordingly which helps in maneuvering safely.

Control:

The last step is controlled, once the car has gathered all the information that it needs it then has to turn the steering wheel and control the throttle and break according to the provided trajectory. To make it simple control theory is a study of how to apply force to an object to control its movements.

So these are the 5 core components on which an autonomous vehicle works. Of course, it is so much more complicated than this but to get the basic understanding this was all that you need to know. I hope you enjoyed this article if you have any queries then make sure to drop down a comment below.

error: Content is protected !!