But then we got to thinking of more realistic co-called ‘Edge Cases’, a phrase common in the Self Driving industry. An ‘Edge Case’ occurs at an extreme – usually
at a maximum or minimum – operating parameter for an engineering system. Now, given that autonomous vehicles ultimately will have no steering wheels, and will therefore be totally dependent
on sensors and AI (Artificial Intelligence) to ensure safe driving over all possible scenarios, this starts to make you think of all sorts of atypical situations that might happen just one
time a year or perhaps once in a million scenarios, but could pose serious safety risks to Autonomous Vehicles. Self-driving cars will have to deal with all sorts of these weird scenarios ….
such as kangaroos jumping across the road in Ohio! Yes, this could just happen as there are probably a few hundred marsupials in the wild across America today having been let loose by owners
and interbreeding into small colonies. And Kangaroos can be very unpredictable animals because they have the unnerving tendency of being in one location at one instant and then 12 feet away
the next – just ask any Australian! This makes kangaroos problematic for self-driving car designers and a perfect ‘edge case’.
Now you know what I mean by ‘edge cases’, you can start to think of other ‘edge cases’ like maybe removals men carrying
a mirror across a street from their truck in front of a self-driving vehicle, or perhaps children dressed up as dinosaurs ‘trick or treating’ at Halloween crossing
the road in front of a self-driving car. Both of these are very rare instances each year, but are genuine ‘edge cases’ nevertheless that AVs will have to deal with in future.
As humans we would recognize and deal with the situation through our innate abilities to judge such scenarios out of experience (or watching movies) and from our learnt memories
as children and make a risk assessment and act accordingly. Somehow, the next generation of self-drive cars being dreamt up today are going to have to deal with these situations
without being paralyzed and creating an unsafe driving scenario. Well, the good news is that VTD can pretty much allow users to put in everything conceivable for visualizing such ‘edge cases’.
Just check out what my colleagues, Karthik Krishnan and Yijun Fan have simulated
as an illustration of a rare problem I do recall from my time in New Hampshire – a not-too-clever bull moose on
the road in front of your vehicle; a major safety hazard as any New Englander will tell you.
|