Coding 10/19/19 (Putting meat on the skeleton)Tags: control
Task: work on actually filling out the auto
As seen in the last post, the skeleton of the auto was done. Tonight My goal was to fill it out-- make it do the things it needs to at the points based on the skeleton. This Would have been a bit more automated had we put a distance sensor on the robot, as I could just tell it to do certain actions based on how far it was from something. Without that, all I could do was hard code the distances. This took most of the time, but was efficient since I did it in stages.Stage 1 - The blocks
My first task was to pick up the first block in the quarry line. I started by going forward and estimating the amount I needed to go, then went into the arm. I needed to make sure that when I went forward, I would go over the block just enough that I didn't move in when moving, but low enough to be able to be picked up with relative ease. So I ran a teleop version of this and got the value for the arm above the block, grabbing the block and just low enough to clear the bridge, a value I'd need later. Then I did trial and error on guessing the distance to the block until the grabber was in position just over the block. Then I ran into a little issue. I wanted to run the intake servos while I put the arm down, but in the StateMachine class, we can only have one action happening at a time per StateMachine object. Therefore, I just set it to run the servos after the arm was put down.
However there was a separate issue concerning that as well. In the intake method, we assign a value to our servo PIVs to control the speed at which they run. This is how you are supposed to do that, the only problem is that that by itself is not compatible with our StateMachine. As we use Lambda functions, the code runs through the lines of .addState() repeatedly until the method call in the method call in that .addState() call returns true. For starters, we had to change the output method to return a boolean value. But isn't it, as if it was left as that, the lambda funtion would always get back a false from that .addState(), and be stuck like that until we stop it. So, I looked back on the old code from last year, and with the help of Mr.V, we found a .addTimedState() method. This takes in a method like a normal .addState() method, but a time to complete can be assigned. With the intake method always returning false, it means that the servos would run until the end of the time set and then it would end that action and move on to the next.
Stage 2 - The deposit
So, after the bock was picked up, the robot was told to turn to the other end of the field, where another set of estimations were used to move forward. This is where the value of just clearing the bridge came to play. To get under the bridge, we have to hold the block and arm in a certain position. After the bridge is cleared, the arm is set to move back up so when we turn to face the build plate, we could deposit. Now this was interesting. As hard as I tried I could not get the deposit to work reliably, but some of the accidental effects gave me ideas for how to get the most efficient way of placing the block. On one of the runs, the block was set down and it didn't quite sit where it needed to, as to say it was tilted back away from the robot. This led to the arm knocking it back into the correct place. I think this is a great way to have a more catch-all way make sure that the first block is correctly placed. I would have expanded on the idea, but I had to leave soon after.Next Steps
I need to test more efficient paths for this auto, but other than that, I just need to finish this version of the auto for the scrimmage.