Talk:DEWBOT VI PARC XIII
Repairs & Modifications Following PARC XIII
The Pivots received a thorough overhaul last night and should be good for Monty Madness this Saturday. We replaced a blown bearing (a spacer had been mis-installed), realigned Transfer Axles and Wheels and tightened all set screws on the Transfer Axles (replacing a couple set screws which went missing). Out of the two swapped-out Pivots, we assembled one working spare Pivot in good condition and ready for use.
The left side Ball Dam Gussett was badly bent in at the bottom. This would have required a considerable impact. Our PARC Pivot failure was on Pivot #3, also in the front left corner, so one hypothesis for the failure is a greater than normal impact to this corner of the robot during PARC Q1. A second hypothesis is that the Pivot which failed was the one with the loosest fit between the Pivot Tube and Pivot Plate (this loose fit was noted during assembly). The bent Ball Dam Gussett was straightened.
A weld inspection revealed no new broken welds. The known broken weld was not repaired.
The 1" Angle frame top elements are all noticeably bent inwards due to competition impacts. Square tubes would have been a better choice (angles were used for weight & Center-of-Mass management).
Latch wire ties were gone. Replaced.
Agreed on IR sensor (2Y0A21 F) mounting position (for sensing that the ball is in kicking position). We'll mount this on Wednesday.
Added side-ramps onto the goal ramp to better match the field geometry (and reduce wear and tear on the robot drive-train).
On Wednesday, we will mount the IR sensor and wire the Accelerometer.
Time permitting, we'll switch more wheel tread mounts over to the Deaver system.
Gary Deaver adds:
- After inspecting the robot Monday night, it is apparent that PARC was a bash fest for us. The pivot top plate tube joint has been a design concern from the beginning. The Pivot tube- top plate joints have held up until last weekend. The joint is loose in only 1 direction. This indicates that the wheel suffered a intense impact. The only way I can see this type off force would be to hit the ramp at just the right angle. This is a driver issue. The drivers need to be careful around the humps and take care with the approach angle. The joint failure is consistent with the simulation that was run in January. The other failure mode was stress cracking at the inner web of the top plate. I have not noticed any fatigue in the top plates. This failure is related to the hardness of the AL. We could go to a 7000 series AL and almost double the hardness of the plates. The problem with this is cost and machinability. I'm satisfied with the current material. Also the box construction of the lower pivot has proven to be more than strong enough. I have seen many other teams with the skinny side plates fail. Ours is robust. Considering the total hours of drive time and the aggressive play I think the pivots have held up well. On a flat field we're good. Any raised field pieces will always be a concern. The window motors are not the proper motor for the job but, it's what we have. The top 1/8" AL angle frame is bent. 8020 quick frame tubing would be a good option. It's 1/6 wall and has a flange for recess mounting the poly covers. 104 built the whole robot frame quick frame and it was durable. Also 1" box pultrusions can be a good choice were impact and durability are a factor. Our pre kick off frame plan had many advantages. Leave it to the GDC to destroy the best laid plans. Last weekend mat have been physical but, next weekend could prove to be the hardest test yet. Some real good teams. As far as the wheel cover attachment, drive and auto take precedent now. Is there time?
Received the 4-wire Accelerometer Cable from Scott Featherman.
Wired the Accelerometer.
Installed and "dry tested" the IR Sensor.
Our Autonomous plan is:
- Line up balls in front of robot and in-line towards goal.
- Set the selector switch to 1, 2 or 3 (n) depending upon how many balls to be kicked (based on starting field).
- Run the Autonomous sequence:
- Align wheels and arm kicker.
- Drive forward slowly (ideally using the accelerometer to keep on a straight path).
- When the IR sensor detects a ball in the correct position kick and then reset kicker. Do not stop driving forward.
- When you've kicked the nth ball, stop.
We also want to use the IR sensor for teleoperated kicking, as an alternative, automatic kicking mode. The drivers would use this kicking mode when working to clear balls hidden behind bumps. When the balls are visible, normal kicking would be employed.
Jon Davis suggested that given our unique drive capabilities, there is no need to drive with the robot front aligned with the field. An improvement on the above sequence would have us align the chassis with the goal using the vision system, then driving straight forward (relative to the field, not to the chassis). This way, we could reasonably score with the balls aligned off the goal axis.
Implementation priority is:
- Get the IR-sensor triggered kicker working correctly
- Get the accelerometer on-line to keep it driving in a straight line
- Davis' goal alignment idea
As of the evening's end, we had no working code to do any of this.
I (Clem) think that a LED/photosensor combination may be more reliable than the IR sensor. When the beam is cut; kick. Will try to test this later, but not a priority.
Noted that the battery box is starting to crack. We make a replacement.
Cut 1/4" thick Tuffak polycarbonate sheet for two replacement Battery Boxes. This time, used a 1" Forstner Bit to produce smooth, radiused interior transitions from the mounting flange to the formed portion of the box. The original Battery Box had a square interior corner and failed at Finger Lakes. Two replacements were made with radiused interior corners, but these were cut using a scroll saw (rather than a hole saw as planned). The scroll saw radii were very erratic and not smooth at all. The first failed in its first match at Drexel, but the second (which had somewhat smoother radii) held up until now. All failures were cracks which initiated at one of the two lower interior corners between the mounting flange and the formed battery box. The rationale of using a smooth radius to correct this problem is to eliminate point of stress concentration at this critical load-bearing flexure point.
The polycarbonate blanks have been put into the oven at 250°F to condition (dry) prior to forming. We will form the Battery Boxes prior to Monty Madness.
We also cut an improved version cRIO Panel, which will protect the compressor and prevent balls from getting pinched between the carpet and compressor when we back into them. Not finished or mounted yet.
Paul worked on autonomous mode.
Two (2) polycarbonate Battery Boxes were thermalformed.
The trailer is packed for Monty Madness tomorrow. All Batteries charged. The 5th Battery Charger was repaired and an apparently unused legal Battery "found". We'll see if this works under fire.