Team Galaxia 5987 in memory of David Zohar is proud to present our fifth robot, Asteria.
Asteria will compete at FIRST Israel District Events 2 and 4 (not as written in the video).
Good luck to all teams, and we’ll see you on the field!
20 Likes
Amazing robot! Outstanding accuracy on a turret and a really cool color wheel mechanism. Good luck this season 5987!
1 Like
Thank you!
What are you doing to keep the climber level? I almost missed it the first time through.
We have two climb mechanisms - one on each side (they are identical). The robot automatically adjusts them in order to stay level.
With the season ending early, I wanted to share some details about our robot here. The team worked extremely hard to make this robot competitive, and despite our unfortunate early exit from Israel 2 I’m very proud of the robot we made. We were excited to make improvements and be back for ISR 4, but that clearly won’t be an option. We took this video in practice the day before we learned our events were cancelled and I want to share it here.
We’ll also be releasing our technical binder at some point in the near future. Meanwhile, here’s a teaser of some cool engineering features on the robot:
“Flipped” turret –
The shooter sits on a turret, with the bearing stacks riding on the outside of a circular plate solidly attached to the robot base. The turret is driven by a custom-machined 125t sprocket.
Pneumatic climber break –
Inspired by 971 and the WCP brake, we developed our own pneumatic friction brake made from a single 1/2" hex hub. The pancake piston presses a rubber brake pad into the climber gearbox side plate, which locks the climber gearbox shaft in place. Each brake was calculated to hold 191 lbs at 60 psi.
Kalman filter localization –
At the beginning of the match the robot uses the targeting camera to determine its position relative to the goal. From this data it calculates the proper autonomous path on-the-fly based on the robot’s true starting location. This allows us to place the robot on the field and let the robot self-align. This localization continues throughout the match using a Kalman filter to combine data from the drive encoders, gyro, IMU, and camera vision to get a constantly-updating field-relative robot position. This position is used throughout the match to help with aiming the turret when the camera can’t see the target.
6 Likes
Wow! We turn out to be doing the exact same thing of using solvepnp for initial position and visual + wheel odometry from that to help with shooter. How accurate is your full field localization? Even with filter, there is significant drift if there is a collision. However, recalibrating the robot pose every time the camera sees the target solves the problem
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.