This is a guest blog post from Alex Van de Weghe who believes for vehicles to be fully autonomous an “organized chaos” algorithm is needed.
Whenever I see some video about autonomous driving cars it’s always in perfect traffic conditions. Not too many other cars, clear traffic signs on the sidewalk and on the pavement. And I’m sure that a current autonomous driving car will stop for any obstacle that come before it and even do an evasive maneuver if the circumstances a right.
When I see a video of how Lidar has improved over the last years, I am impressed. The depth precision, the color reading of the environment and all this in real time makes it possible for a number of vendor programs to really analyze what’s happening around the vehicle. So, we can say that “seeing” around the car isn’t the biggest problem anymore.
The currently available autonomous driving software has embedded several levels of safety. Zero being just drive and stop for a serious problem and five being on the highest alertness for any possible obstacle, plus give priority to pedestrians, animals and bikers. Some would argue that it’s clear that the software is almost ready to operate under a range of circumstances.
But when I look at traffic situations all around where I have travelled (Europe, Africa, Asia) I know that if an autonomous driving car on level five would be dropped in the traffic in a major Asian city or in an Italian city with the very narrow streets, it probably would not even move.
There are simply too many variables such as scooters, street lanes that are stopping or changing without any predictable reason. It is likely that the software would simply direct the driver to take over because the safety cannot be guaranteed. Operating with the software on level three is not the solution to this problem.
I believe that what I call an “Organized Chaos” algorithm is needed. An algorithm that calculates not only how the traffic is moving and how narrow a traffic lane is becoming, but at the same time calculates the enforcement of how to proceed and the factor of safety if the car would follow the path that would lead it through the traffic. We could call it the law of the enforcer, which is often how cars are moving through the traffic in some parts of the world without collision. And it works, because everyone is moving (driving) that way.
That same algorithm needs to calculate that if the traffic moves in a very orderly way, with a very clear structure that this level of enforcement needs to be almost zero to follow the non-chaotic laws of traffic movement. This is because it also works that way.
To be clear, this has nothing to do with the safety level. It still needs to stop for any obstacle that could lead to a collision, but I believe that the one can’t properly function without the other if autonomous driving is to become the future.
Note – If you liked this post click here to stay informed of all of the 3D laser scanning, geomatics, UAS, autonomous vehicle, Lidar News and more. If you have an informative 3D video that you would like us to promote, please forward to firstname.lastname@example.org and if you would like to join the Younger Geospatial Professional movement click here.