During my spare time I retrofitted one of my robots with a customized paintball-ejecting accessory. I installed the system onto my largest robot, a Fanuc S-420 series, which is located adjacent to a loading door facing the main building. The robot is part of an integrated workcell that also consists of a CNC lathe, an industrial monitoring computer (PLC), and a Vision tracking system that is normally used for part location verifying. The robot is capable of visually tracking an object within a certain distance. I call this entire system the ROBO-ION.
The marker is powered using a DM4/5/C circuit board that I interfaced to the robot's signal interface bank.
Timing of the firing sequence is determined by the industrial monitoring computer (PLC). The PLC communicates with the robot, which then selects from pre-defined pulse settings in order actuate the marker's solenoid. This signal varies depending on the desired firing mode (semiauto, ramping, etc). Normally I would simply have the robot control the marker's firing directly, but I chose not to do this since the Tadao DM board can excert better control of the millisecond timing compared to the robot.
The robot determines target position based on an adjustable coordinate system used in other vision-related activities. This adjustable coordinate system is typically used by the robot's motion processor as a way to calibrate the robot to different movement positions based on other obstacles inside the workcell (tables, machines, etc). This allows quick and easy adjustment of the firing range, or KILLZONE as I like to call it. The killzone isn't perfect, but then again it combines coordinate tracking via third-person camera system, combined with a laser range sensor mounted to the robot, combined with the already-inaccurate paintball firing, so it's impossible to achieve a perfect system just based on the limitations of the hardware itself. For example, the lighting in the room plays a pretty heavy role, and the system won't work at dusk or nighttime.
Tests 1, 2, 3 - Stationary Target
I made these videos while editing the coordinate systems. Fanuc's Vision adjustment already has built-in models for visually tracking bottles and cans, so we used those for testing.
Sorry, I ran out of air at the end.
Here's a joke video I made, where I pit the robot's aim against that of my associates. The robot's aim is much better, though it takes time to track the objects and settle into position. Interesting results...
Test 4 - Moving target(s)
Coming soon! (if I put the Ion end-of-arm tool back onto the robot)