Recently, we visited the University of Michigan's Mcity, which is a testing ground for autonomous and connected vehicles. This testing ground features almost every imaginable city driving situation in just 32 acres. Actually seeing the facility, and considering the sheer number of object and road variations on display, Mcity illustrates just how impressive a task it was to fit this all into a tiny space. It also shows how complicated the environments that autonomous cars will face are.
Most of this complexity comes from how many of the everyday objects we see have subtle variations that are important for autonomous vehicles to take into consideration. Take traffic lights which can be hung in different ways, such as strung on wires or mounted on poles. Then there are different road markings. They can be narrow, wide, and have parking spots. And those parking spots which can be parallel, perpendicular or angled. The roads can even be marked to simulate European roads, which is a whole other set of complications. Then there are different buildings, building placement, street objects and more. There are even different kinds of curbs. Seven to be exact. Did you know there were seven kinds of curbs used in cities? We didn't. But Mcity contains all these different variations.
These are complex issues in ideal conditions, but the world isn't ideal. So Mcity also includes sub-par road scenarios. The highway section is part concrete and part asphalt to make reading lanes more difficult. Worn-out and vandalized signs are scattered about. There's a stop sign that isn't red anymore, and a pedestrian crossing sign with stickers all over it. The city also features a corrugated steel tunnel designed to simulate how a tunnel or overpass could block signals, a hill designed to test vehicle-to-vehicle communication, and a canopy that simulates tree covering. What effect do trees have? Well apparently they can block GPS signals.
This microcosm of a city illustrates just how incredibly complicated something as mundane as your hometown really is, and we haven't even covered everything. If you want to see more, you can check out this video from University of Michigan.
What's really amazing is how we, as humans, are able to process all of this without thinking about it. We see a traffic light, and we don't freak out because it's mounted differently. But these are things that have to be programmed into computers, and the computers have to be able to prioritize what information is good when some sensors start going haywire. And all of that is why a comprehensive test ground such as Mcity is very important to making sure the computers can cope with our world.
Related Video:
Most of this complexity comes from how many of the everyday objects we see have subtle variations that are important for autonomous vehicles to take into consideration. Take traffic lights which can be hung in different ways, such as strung on wires or mounted on poles. Then there are different road markings. They can be narrow, wide, and have parking spots. And those parking spots which can be parallel, perpendicular or angled. The roads can even be marked to simulate European roads, which is a whole other set of complications. Then there are different buildings, building placement, street objects and more. There are even different kinds of curbs. Seven to be exact. Did you know there were seven kinds of curbs used in cities? We didn't. But Mcity contains all these different variations.
These are complex issues in ideal conditions, but the world isn't ideal. So Mcity also includes sub-par road scenarios. The highway section is part concrete and part asphalt to make reading lanes more difficult. Worn-out and vandalized signs are scattered about. There's a stop sign that isn't red anymore, and a pedestrian crossing sign with stickers all over it. The city also features a corrugated steel tunnel designed to simulate how a tunnel or overpass could block signals, a hill designed to test vehicle-to-vehicle communication, and a canopy that simulates tree covering. What effect do trees have? Well apparently they can block GPS signals.
This microcosm of a city illustrates just how incredibly complicated something as mundane as your hometown really is, and we haven't even covered everything. If you want to see more, you can check out this video from University of Michigan.
What's really amazing is how we, as humans, are able to process all of this without thinking about it. We see a traffic light, and we don't freak out because it's mounted differently. But these are things that have to be programmed into computers, and the computers have to be able to prioritize what information is good when some sensors start going haywire. And all of that is why a comprehensive test ground such as Mcity is very important to making sure the computers can cope with our world.
Related Video:
Featured Gallery University of Michigan Mcity
Sign in to post
Please sign in to leave a comment.
Continue