Stanley was not in this race, it was in the 2005 Grand Challenge and won that race. This one required legal driving in an urban environment, ie, obey traffic laws, stay in the proper lane, yield right of way, etc. All of the vehicles were given the same map info, how they chose to use the data was up to each team. This visualization shows the legal turns that can be made at each intersection, that is how it knows where to drive.
This might work, but this method requires incredibly detailed maps. The way stanley works is it actually learns what to look for. Now if the car can actually generate these maps, there is indeed hope for this car. This is still incredibly impressive stuff.
Stanley was not in this race, it was in the 2005 Grand Challenge and won that race. This one required legal driving in an urban environment, ie, obey traffic laws, stay in the proper lane, yield right of way, etc. All of the vehicles were given the same map info, how they chose to use the data was up to each team. This visualization shows the legal turns that can be made at each intersection, that is how it knows where to drive.
leenstl 2 years ago
This might work, but this method requires incredibly detailed maps. The way stanley works is it actually learns what to look for. Now if the car can actually generate these maps, there is indeed hope for this car. This is still incredibly impressive stuff.
ragonamuffin 3 years ago
WOW Nice software!!
But what about the green vs red obstacles? Is it OK to drive over the green ones? :D
batukhan 4 years ago