1678’s robot has a green light that appears to be for vision tracking. This light flashes during the time I assume the turret is tracking the goal. Why is that?
I have two theories:
- The light is off for one camera frame and on for the next. The vision algorithm compares the two frames and uses it to filter out only the pixels that change. I assume this avoids interference and reflection from other tracking lights on the field.
- It’s just a signal to the driver when the turret locks on or the shooter is up to speed or something.
Does anyone know the answer? @Michael_Corsetto?
3 Likes
which light are you talking about? They have actual limelight green light, but they also have signal light on their turret that turns blue, red, yellow, and green(If im remembering correctly)
We’ve thought about doing something similar but with RGB LEDs and changing the colors.
As far as I can tell, there’s no limelight on their robot. The only lights I can see in the videos are the ones on top of their turret hood. They do change color, but I was assuming they were used for tracking in addition to signaling.
1 Like
They do have limelight on their robot(I volunteered at LAN and saw their robot) and they use the rgb to signal whether the targeting was successful or still in need of adjustment.
Ah okay, I see. Thanks for the clarification!
1 Like
There is a limelight that’s used for targeting. The LED strip functions as a signal to the driver/operator as to what’s going on. The blinking green means that the robot believes it is “locked on” to a target. The robot stores target information as field relative timestamped coordinates, so the lights may be blinking green even if the robot can’t actually see a target but is aimed from odometry data. This is imperative as our limelight is positioned above our shooter and the camera is blocked for most of the shot by balls flying out of the shooter.
The LEDs are also used on disable to signify calibration, and during the climb sequence to signify climb states.
16 Likes
The limelight is positioned on the hood at the same location as the LEDs
3 Likes
I just need to know if they’re actually using 4 (no that’s not a typo) falcons on their shooter.
7 Likes
Yes.
Two Falcon 500’s power our primary (4") and secondary (2") shooter wheels.
One Falcon 500 controls our shooter hood.
One Falcon 500 powers our feeder omni wheels that suck in the balls from the spindexer.
Best,
-Mike
16 Likes
This is super cool. How accurate is the odometry data usually?
4 Likes
What ratios are you using we ran into the same problem as everyone else this weekend with the poor quality of balls effecting our shot. Didnt seem to affect you guys near as much. Just looking for a place to start on the redesign more compression or more speed.
It doesn’t drift by more than 1-2 degrees during normal driving for the first 5-10ish seconds, but matchplay skidding + colsons can easily cause 3-5 degrees of drift by the next aiming sequence. Ultimately, the goal at that point is to just keep the turret and hood close enough to where they can aim after untucking the hood without operator intervention.
9 Likes
Do you reset odometry positon at all during teleop? Such as using limelight distance calculation(although can be noisy) or just shooting in a known location?
I realize I may have been a bit unclear. The odometry is used for keeping the turret aimed from aim sequence to aim sequence. So, the robot doesn’t really care where it is on the field, what it cares about is how much on the coordinate system it has moved since it last saw the target to estimate where the target is relative to its current position. Thus, it’s unnecessary to be field-accurate, but the target-accurate odometry is effectively reset every time it sees the target at a certain stability.
8 Likes
Are the 2" shooter wheels 1:1 with the 4" shooter wheels or do they spin at a slower RPM?
The 2" wheel is geared to be ~60% (24:40) the speed of the 4" wheel, which gives ~30% the surface speed.
4 Likes