We feature many great Kickstarter projects on Geekboy and this next one is no exception. The project, a collaboration between Charmed Labs and Carnegie Mellon, is Pixy, a super-fast vision sensor which can be taught to find pretty much any object that you want it to.
Most image sensors output huge amounts of data and this can be overwhelming to a lot of processors, hindering the processors ability to deal with other tasks it may be needed to do.
Pixy has its own dedicated processor and it only transfers the necessary information to your Arduino, or other microcontroller of choice, sending it at a rate of 50 frames per second. This takes some of the burden away from your microcontroller, leaving more processing power available for your other jobs.
The latter algorithm means Pixy is able to identify hundreds of objects at a time.
The colour algorithm enables it to remember up to 7 different colours and with colour coding (two or more colour tags placed close together) it can remember many more colour combinations.
The clever device can be taught to find whatever object you instruct it to, by placing it in front of its lens and holding down the button on its top. It has a little RGB LED at the front which mimics the objects colour when it has fed back the visual information necessary. I.e, if you show it something red, the LED will turn red, then it will look for any items with similar colour signatures.
The data obtained can be accessed via several interfaces including : UART serial, SPI, I2C, digital out, or analog out. It’s clear that the team intend for pixy to be as accessible as possible!
This is a great project to get involved in, with the added bonus of an application that you can run on your PC or Mac which enables you to see what Pixy sees. The application is called PixyMon.
This with everything else in this campaign is open-source, meaning it is open to build on, change or do whatever you want with it. Maybe you will create your own custom Pixy project!
When you consider that the typical tracking projects involve a camera and often a PC to process the huge volume of image data, its easy to see the genius in this device.
Earlier in the year we wrote an article about a tracking Portal turret using Arduino and a PC. This effectively achieves the same goal but with much more hardware / cost.
It will be interesting to see how this technology evolves and even more so the projects it inspires. Quadcopters… What do you think?
Head over to Kickstarter for more details on this great project.