Autonomous cars all the craze now a days and there is lot of progress being made with the underlying tech. There are still areas tech cannot help for example, making ethical decisions like: if it comes to that point, who do you save? The pedestrian or the passenger? All this has to be resolved via democratic/policy decisions and am sure will be taken care of once the technology itself matures.
Now, getting back to technology, there is this thing I came across, that is, the algorithms are having a hard time recognizing the bikes on the road. If we break it down, the issues seem to be :
- First they are small and they weave unpredictably
- They have a general shape but it still vary widely based on the viewing angle, rider and luggage
- The direction of travel of the bike
- Sometimes they come from nowhere!
These seem to be issues at this moment but I believe, are within the reach of a clever complex algorithm, except the last one. It is going to be impossible to suddenly react to something that appears in your view! There are some folks at work on these issues but as of now it is a #ProWoSo!
Sorry to hear the loss of bikers’ lives (in recent news) at least the ones that could have been prevented, but were lost because of un-attentive observer-drivers!
[Image Credit: Thanks to Pexels via Pixabay]
Why would direction of travel be an issue?
Because if the bike is moving toward you, the car might have to make a different action compared to when the bike is moving in the same direction as the car.
Dear Sir, I am doing work in AI. Is this problem solved? I cannot google any references, please help me.
-Kapil Sinha