At a recent Sturdy Hack-a-thon we utilized Amazon Rekognition along with a stock Raspberry Pi3 to customize a NERF Rapid-Strike gun for a fun solution to an office problem, a busy CTO who’s sometimes a little distracted.

Choosing the correct gun for this project was important, as our requirements were that it were fully automatic in operation and could be actuated electrically. We ultimately choose the NERF CS-18 N-Strike Elite Rapidstrike as it met our criteria and is mod-friendly. The gun needed to still operate normally while not being controlled, needed to be able to be powered even with batteries installed, and should be able to be “overpowered” to shoot faster and further.

The controller perched atop the gun is the “brains” for STING, controlled by a stock Raspberry P1 3 and Amazon Rekognition. The CPU was chosen for its ease of use and available software libraries, and then partnered with an Adafruit 2.8” PiTFT display with Adafruit enclosure and Faceplate. The camera used is the standard Pi (version 1.3) with an Adafruit Pi Camera Board Case.

For our particular dilemma, we used Rekognition to add all team members to a “safe” list that will hold fire, while our CTO was added as the target which means a barrage of foam projectiles once STING identifies him.