We’ve finally managed to get our catapult mounted to our drive system and implement the first stages of the controls to make it all work together.
We have an unofficial motto – fail often and fail fast – that drives our design process. The idea is that we learn more from failure than from success and, by rapidly testing ideas and seeing how things perform, we’ll build a better robot in the end.
So, in that spirit, the linked video is about 5:30 minutes of unedited footage showing some of the first tests of the system as it exists today. Maybe it’s not exciting as a sizzle reel of awesome robot feats, but it shows our real progress – not selectively edited shots showing only successes.
The automated aiming and range-finding processes we’ve used in the past – in 2010, notably – are not implemented yet here. The aiming and distance adjustment is being performed manually and so they’re pretty much guessing at the correct values.
We are hoping to get away with aiming using just the drive – an 8WD system with two center wheels dropped and a resultant 7.25" wheelbase.
Distance is regulated by controlling both the pressure to the pneumatic actuators and the flow rate of their exhaust. This allows us to vary power and arc independently.
In the end, we expect to automate the entire process of shooting balls, but we’ll need the camera set up for that as well as some time to dial in the correct pressure and flow settings for a set of distances.
This was only a test of the catapult with the drive and pressure/flow regulation systems. The ball collector, ball storage and catapult loader are yet to be integrated onto the robot.
May I ask? I can’t tell from looking at the video. Which of the following describes how your catapult works:
In the “ready-to-fire” position, the “retract” side of the cylinder is pre-pressurized and the “extend” side is vented, and the arm is held in check by a latch of some sort. The latch is released to fire. Then the “retract” side is vented and the “extend” side is pressurized to return the arm to the “ready-to-fire” position.
or
Same as 1, except their is a spring (or surgical tubing etc) to assist the launch.
or
Same as 1, except there is no latch, and both sides of the cylinder are vented in the “ready-to-fire” position.
I’m not sure if I can explain how it works properly – pneumatics is not an area of expertise for me – but I’ll try.
We have a single valve acting as a gateway to what basically constitutes a second pneumatic system for fire control. That valve has a pressure sensor in line with its output, several accumulators and one additional valve for each of two firing cylinders – so, two more valves there.
We set a desired pressure value for this system to determine firing distance, open the gateway valve and let the accumulators fill to the desired pressure. When we’re ready to fire, we open the fire control valves to move the two cylinders that actuate the throw.
There is a ball valve attached to the exhaust side of the throwing cylinders that can be adjusted via servo. We can meter flow rate here to further affect the ball trajectory.
The whole system is also connected back into the main 60psi line with a check valve so that our air dump valve clears everything out as required.
Thought I might add some more video of the catapult testing. We wanted to demonstrate scoring in a more realistic environment – so we prioritized being able to drive the robot around and score reliably; no use setting it in one place and making sure conditions are perfect.
This video shows automated range-finding and firing. Aiming is still achieved manually here with the help of the camera’s video feed to the driver station, though the robot is itself calculating range and setting up our firing system appropriately.
Shortly after this video, but not before my battery died, we were able to achieve success with automated aiming as well.
Also, if you’re interested, here’s video of the first end-to-end test of the robot. We hacked some things together to see this work and the loading mechanism is terrible. That’s been fixed.