This is the first of a three-part series study of the engineering behind self-driving cars (autonomous cars or robot cars) taking Waymo as the case study.
At the end of this series, you may not be able to build your own self-driving car, but you would surely understand it enough to start an enlightened conversation on it.
In this part we are going to be discussing about the ground knowledge surrounding self-driving cars
Let’s begin…it is autonomous not automated
Any talk about self-driving car, in my opinion, should begin with a fundamental understanding of the difference between autonomous and automated.
Self-driving cars are autonomous meaning they are self-governing and can act alone, while automated (from the word automatic) means that its control is more mechanical.
Autonomous is what we would discuss and not automatic.
Autonomous is what we want to see in the future. Automated car is what we are driving today – emergency brake, autopilot in some cases. . .
So, when is it autonomous?
From the start of research on autonomous cars (self-driving cars), there has been various definition. But we still all agree on the classification of the SAE (Society of Automotive Engineers) to help us define automation in cars.
And the levels of automation are:
- Level 0: These are automated systems issuing warning and may momentarily intervene. It is present in most modern vehicles e.g. emergency brakes, fuel level indicator etc.
- Level 1: At this point the driver and the system share control of the vehicle. It is the hands-on level.
- Level 2: In this level, the driver can “hands-off” and should be ready to intervene when necessary.
- Level 3: Now, the driver can take his “eyes-off” the road
- Level 4: This level is when the driver can take his “mind-off” the driving. He/she could probably go to sleep.
- Level 5: This is the apex of self-driving cars, when no intervention is required.
So, if is level 3 or greater, then it is autonomous…get it
One thing before we move on…
As of the time of this writing, only the Google self-driving car or Waymo has been reported to reach level 4 on the experimental level (controlled environment). But there are a lot of doubt when it would reach such level on the streets.
And that’s why we are going to be talking about Waymo, but not only that also because of these:
- It is been developed by Google which indicates they have almost unlimited resources to achieve the feat.
- It has the most distance covered on its autonomous system (8,252.6 km/ 5,127.9 miles).
- It has the most successful and promising future until this moment (out of all (3 maybe) accidents with the car only one has been caused by the car itself, the rest is human error).
- It is now within reach to enthusiasts who want to enjoy the self-driving car experience
- They have spent $1.1 Billion between 2009 and 2015 on self-driving Tech
So, who is Waymo?
Waymo’s journey started in 2009 as the google self-driving project under the supervision of Segrey Bin (Google Co-Founder) in their (Google) secretive division, X-Lab.
Then there came Alphabet, a parent company that was founded by google founders to house all of Google’s ecosystem. It was then it became a subsidiary of Alphabet and took up the name “Waymo” as a separate company.
In 2017, they moved to an Old Airforce Base, Atwater, California US for their testing and development. Though there are still some sections in the Google Headquarter that is deemed for Waymo, they are fully operational at the Base.
I mean, what’s involved in the design of Waymo?
Any self-driving car is composed of two parts, the hardware (car propulsion system, frame, sensors, actuators etc.) and the software.
With the exclusion of the design of a car which I would believe they didn’t need to redesign, we would talk about the software technologies and the hardware technologies that makes it self-driving.
It practically building a brain…yeah, a brain.
The google self-driving car like any other self-driving car needs to develop a brain. And the brain is developed in a series of testing defined by the sequence describe below:
• Get data from the environment
• Use the data to determine the right controls
• Send the control signal to the actuators that control the output
Simple, right? Naha..
And that wraps up the first part of the series.
Key things to remember:
- Autonomous is not automation
- Only levels greater than 3 are autonomous vehicles according to SAE
- Waymo is by far the most successful of all self-driving cars to the moment of this writing
- Waymo is owned by Alphabet which also owns google, Youtube, Deepmind, Go’s product etc. Hence, they have the resources to pull it (self-driving cars) off.
In the next part we would talk about the sequence of operation in detail and how it helps them achieve autonomous in their cars. Stay with me.