Test Beds

Hey all,

In developing my own team’s control system test bed, I was curious what your teams do in order to test code or components off of a robot. Do you use WPIlib’s simulation capabilities? What does your physical test bed look like? How do you utilize it during and outside the season? Pictures, tips and tricks, and other general tidbits are more than welcome. I’ll include a picture of our WIP to start.

1 Like

Here’s a sneak peek at 696’s testbed rack (I’ll put together a whitepaper at some point):

The bottom four computers are unrelated to the testbed, they’re just CAD render nodes.

11 Likes

We also have a portable test bed that we like to call “Robot in a box” It is comprised of the entire control system plus a few “Axis” to control. One of the control “Axis” is a leadscrew which driven by a motor/encoder setup via a Talon SRX. The other “Axis” is a flywheel setup. This allows us to train our programmers on configuring/tuning/testing a positional control and velocity control. Also since its portable the different programming students can take it home to continue their development work.

I wish my team had something like these so If anyone has plans for theirs I would love to see them

We use a basic testbed almost identical to yours, 21brownz. We tend to attach whatever we need to the basic control system in order to run tests (motor controllers, solenoids, etc.) and then attach the systems we need to test to those. It’s not overly sophisticated (unlike 696’s setup, which is somewhere in the “holy mackeral” zone) but it gets the job done 99.99% of the time.

BTW, looking forward to seeing you guys at THOR on Saturday.

If you are just looking to just test a prototype mechanism (say a flywheel shooter) and not test the code for that mechanism, you can bypass the code by driving motor controllers with PWM signals from a servo tester like this one or by using RC radio gear.

See you Saturday.

1 Like

What advantages does having your test bed be rack mounted provide? Does the lack of portability ever hurt you? Can you explain a typical use scenario?

Is power still provided via a standard FRC battery, or are you using a AC/DC converter to provide power via an outlet on the rack?

2 Likes

3946’s test bed sometimes looks like a previous year’s robot or practice robot, sometimes just like a piece of board [e.g. pegboard] with components tie-wrapped down.

One of 3468’s goals this year is to make a test bed, since we’ll get back a spare RoboRio and parts from switching the control system of our Air Cannon to Arduino. So thanks for the inspirational photos!

19-inch racks are a fantastic solution for keeping a bunch of hardware together in a compact footprint. It takes up a lot less floorspace in the electronics room than a robot, and can’t slam into a wall or ceiling if there’s a bug in the code.

Not so far.

Power comes from a programmable power supply-- currently an HP 6632a, but we’re looking to upgrade to something like a Keysight N6701C.

The concept is to be able to run automated hardware-in-the-loop tests, and subject the code to any situation it might encounter in competition. So, for example, a test script can deploy code to the RIO, enable it, and then command the power supply to dip to 6 V to test the code’s undervoltage response. We have one of each motor controller that we use and a CANbus monitor to troubleshoot communications. We also have a server (not shown above-- those are old pictures) running eSXI, so we have a VM with windows to run the driver station, an Ubuntu VM with Jetpack to play with the Jetson, and so on. My next project will be setting up something like netem or wanem on a VM and routing DS-RIO traffic through it, to simulate WiFi issues and dropped packets.

1 Like

here you go!