We were chasing a communication issue all day Friday at Atlanta. Our robot randomly lost connection during 3 of the 5 matches. The modem would have a blinking green security light and would would work fine once we reset power. We tried a new radio and new ethernet cables, none of which seemed to help.
Saturday morning we realized that due to how we mounted the radio (in a corner with face against our frame), a hard bump would cause the button on the front panel to be pressed, resulting in the radio trying to automatically connect to a network.
Lessons learned, Murphy’s law is still in effect, and don’t mount the radio where that button can be easily hit.
Definitely. We didn’t have the button issue, but we did have the radio mounted right under a drive motor, so we were getting a lot of interference that we didn’t know the cause of until our 3rd qualifier.
In one match we had a no start problem in tele after moving OK for autonomous. WE did a reset of the DS and started to move ok for rest of match.
In another match The above scenario happened at the start of autonomous.
Much later we discovered it was an intermittent radio to cRIO cable that caused this. The robot likely started not because of the DS reset, but rather a concurrent hit from another bot as game play started. Our Radio - cRIO Ethernet cable was not always strain received at both ends and I suspect that the constant flexing from hits weakening the crimped connection. We will seriously strain relieve all cables next year. We will also shock mount the radio on a damped (non oscilatory) mount. Thhe bot radio is a consumer radio and is not really designed for hard mounting on a bot that takes hard hits.
We also found in one match (we lost) where the bot did not start and the radio yellow light was flashing indicating it was trying to resync. Field staff told us bad radio and we got another one.
Then in our last QUAL match, the first with the new radio, we started OK but had a bizarre control problem that cost our alliance the match. Driver had low gain forward and backward control and limited turning radius. Also, the trigger did not work on either joystick (we could not even shoot our initial load of 7 balls). The button for our conveyor did not work either. i.e. no joystick buttons worked. Our front wheels (independent steering loops) were synchronously dithering at 2 to 3 Hz with a p-p amplitude of ~45 degrees. Notice I did not say oscillating. For the synchronized dither movement to take place both loops need a setpoint change or some other common denominator. If it was control loop oscillation I would have expected independent oscillatory movement from each front wheel.
It almost seemed like the steering loops were being commanded the forward ( neutral) position between alternating good setpoint packets and the buttons related packets did not make it at all. WE did not do a DS reset this time because we did not really discover this until about 15 to 20 sec into the match and then they jiggled DS cables for another 10 sec. The DS reset would have taken us out totally for another ~30 sec so we decided to soldier through the match with limited control instead of possibly none at all. In hindsight (because we lost) the risk of not starting again and being immobile for the reset would have been a worthwhile gamble.