Premise:
The first idea is a way to improve on Global Positioning System (GPS) navigation for cars. The way that GPS navigation works is by using satellite signals to figure out where a GPS device is and calculating data on its movement, generally just speed and direction. This works fine for me, until I hit less-than-ideal conditions. Other times, I’ve driven along regular stretches of road and had my device fail to find a signal, leaving me without directions. I’m not sure whether it’s just a bad unit or there are actually problems with the GPS “network,” but this got me thinking about workarounds to the issue I have.
The concept:
Replace or augment the current technology in navigation devices with a camera-based system. The navigator is placed on the dashboard, and a connected camera looks outside the windshield. Elements of what it “sees” (signs, buildings, trees, and so on) are compared to a database of pictures from around the world. The device accurately pinpoints its location based on the things the camera sees and where these things suggest the device is. For example, the camera sees a Starbucks logo somewhere on the left side with a tree behind it, a single yellow line on the pavement (meaning it’s a two-lane road), and a gas station on the right. Using technology like a face-recognition system for landscapes, the device compares the image to the ones it’s indexed. It quickly finds a match and determines the device’s location based on that match.
Besides this, navigation from point A to point B is pretty much the same. The device figures out the fastest way from one place to another based on the built-in map data it has (which should be identical to the map data on the GPS systems people have today, but separate from the pictures in the maps).
The pros:
Google Maps’ Street View feature as well as image services with geotagging have made the foundations for a database of images that systems can match against what they see.
No satellites to search for means constant access to accurate location.
This technology does not necessarily require new hardware. An iPhone app may be the most practical proving ground for a developer.
By crowdsourcing maps, current pictures taken by the navigation devices can allow easy updates and natural growth for a system. If a device is in an area with pictures that don’t quite match the indexed ones in the database, then the user can upload his/her device’s pictures to replace the database’s old ones.
In a similar vein, road map updates (and weather and traffic updates) are made easier and faster. Pictures from devices confirming that there is road construction, for example, signal that there is an issue with with the current map. “Seeing” snow and a large number of cars is an organic way to confirm the driving conditions, which are snowy and congested in this case.
Reporting aggressive drivers may be done more accurately, and possibly even automatically.
Counting the number of users who pass a billboard may be useful to the advertising industry’s media and metrics sectors. Collaboration with marketers may open up new possibilities.
The cons:
I’m guessing that it would take a lot of processing power to figure out exactly where a device is, especially at night. This technology is likely a long ways off into the future. It would likely start with an in-the-cloud approach at first, followed later by computations being performed on the devices themselves.
Until pictures from all over the world can be stored on one device and updated at short intervals, this system could run into the same connection issues as GPS. Those pictures have to be stored on a server, then uploaded and processed live by the navigating hardware. The result then needs to be sent to the device. If there is ever no connection between the navigation hardware and the server, then the device is crippled. On the other hand, it may be possible to make up for this at times with a built-in GPS.
Landscapes are constantly changing, so each location in the database needs to have pictures for various weather conditions. Additionally, updates must be made more frequently during times of construction in an area.
There may be a few glitches if different places have similar features.
Like with the last one, a picture dilemma may exist with long stretches of highway. On the plus side, a built-in memory may be useful. If it can remember that it most recently passed mile marker 52.9 on highway X, the navigator can be relatively precise in its location-finding.
If the system operates with still pictures taken at intervals (as opposed to a constant video), the built-in camera must have a high shutter speed. Otherwise, blurring will occur in images taken when the vehicle is travelling quickly.
No comments:
Post a Comment