Virtual Test Drive & Autonomous Driving

Autonomous vehicles are becoming more and more realistic by time. Thousands of vehicles are being tested on the roads by companies such as Tesla, Uber, Waymo, Cruise, Didi, etc. Almost all of these companies have accumulated millions of miles of road-testing data, which in result have helped them enhance and validate their respective algorithms.

Just like every other vehicle, autonomous vehicles should also undergo almost every traditional testing and quality assurance. On top of that, it should also prove to be as much as safe or reliable as a human operator or better.

So, the algorithms should inevitably should prove the same, for it to be let out into the world. But most regulations will not allow these as it is not that safe yet - even for testing. Although now some cities have allowed certain companies to do the testing on the road, that too, in limited areas and in limited capacity, that simply won’t present much scenarios for the algorithms to learn.

The virtual testing environments have now the capability to solve much of these problems. That is why most of the top players are pursuing the same. But what solutions does the Virtual Test Drive (VTD) offer and how can they be leveraged to our current needs.

Let’s go through some of them.

1. Cost Saving

A fully equipped autonomous vehicle may cost around half a million dollars. Thus, making the cost for a small fleet of 20 cost around 10-12 million dollars of investment, just for the hardware. There are a lot to go still on regulations and approvals and then finally testing on road. In case of VTD, you only need to create a base model and then make variations accordingly in order to test it on virtual environments. You will still have to test it physically on the road though. But now, the algorithm would have learned so much from the virtual environments before it even hits the road. Money will be saved by reducing number of prototypes in effect.

2. Distance Coverage

Back in 2018, the commonly accepted number of 1 Billion miles was supposed to be required for developing an autonomous driving system. By January 2020, Waymo had covered around 20 million miles on public roads, but around 7 Billion in virtual environments, since its inception back in 2009. Today that number is over 10 Billion miles. Now, if they had only relied on the testing on public roads and followed through with the same rate, it would have taken around roughly them around 5.5 centuries to reach the 1st Billion.

3. More Scenarios to learn from

We could test them on real roads, if you only need to check a few use cases. However, we could never assure the use cases we encounter in real roads by chance would be enough for the autonomous vehicles to be fully ready. The number of conditions to be checked will scale quickly to millions and there is no way to meet these needs without virtual environments. For instance, what happens if a city decides to paint all the road signs yellow instead of white? What happens if trees grew larger to block the view from seeing the pedestrians? What of children dressed as pumpkins out for a walk in Halloween? These scenarios are to be considered but might not be encountered in the real world.

4. Safety

There is a high probability of the vehicle not detecting a person or an animal and having a head on collision while testing. There is a chance of hitting other vehicles on the road or parked on the side of the road. There is a chance of not recognising a traffic sign and meeting with an accident. The vehicle might go out of the lane or even the road. All these scenarios will still occur in the virtual environment. But the differences are, there won’t be any harm to any person or animal and no lasting damage to the vehicle prototype or third-party property.

As more and more companies are leaning into Virtual Test Driving, this list could only increase and contribute more for a driver-less future.

Disclaimer : The views and opinions expressed in the article belong solely to the author, and not necessarily to the author's employer, organisation, committee or other group or individual.