In 2012, the FIRST challenge involved playing 3 on 3 basketball and balancing on a bridge, in short. Here’s the game animation for a better overview:
In 6 weeks, our team was able to design and build our 2012 robot named “Triangulus Prime GT:RS Superleggera” after its side profile and excessively high speed in comparison to the 2011 robot, Triangulation, and in general.
- Coaxial swerve drive geared for 8 and 21 feet per second (2.4 and 6.4 metres per second)
- Drive modules were driven by vertically mounted Andymark Supershifters using servos to shift
- Power is transmitted through chain linking 2 modules together, front and back, then miter gears, into chain, to the wheel
- The frame of the robot is bent to aid in turning, adding caster to all 4 wheels
- Encoders track steering angle and wheel position
- Drive modules can rotate continuously without harming anything.
- 3 sided ball pickup
- 2 side conveyors driven with round belts feed into a center conveyor
- A single longitudinal conveyor runs the length of the robot and feeds into a magazine for the shooter
- Intake rollers were made of electrical conduit with dead axle “stubs” to minimize weight
- A single arm bridge manipulator
- Powered by the robot’s drive motors in a “power take off” arrangement
- Proximity sensors and an encoder allowed for the robot to detect the presence of the bridge and lower the arm appropriately
- A 2 wheeled shooter, later changed to 1 wheel
- 6″ rollers spinning at up to 5000 rpm compressed and launched foam basketballs on contact
Reveal video courtesy of Chris Choi:
A video to illustrate just how fast high gear was, courtesy of Kevin Ho:
The justification for using a swerve drive in 2012 was being able to drive on to the bridge sideways, allowing for a triple balance and 40 bonus points. This at least worked, and was the first in Canada at UOIT:
The balancing process starts at round 2:20:
In retrospect, the robot didn’t perform well for a number of reasons, but we’ll start with the drive train I was behind. It was intended to be lighter, more robust, and faster than ever. Linking drive pods together was intended to neutralize “torque steer” from the drive which is a common problem on coaxial swerve drives (the same shaft which drives sends drive to the wheel is the same shaft modules are steered on). If one wheel were to lock up, a steering motor would usually have to hold its position for a robot to push, but in on our robot, the tension would be taken by the steering chain linking 2 modules, relieving the steering motor of any load.
Here are some sectional views for a better idea:
Unfortunately somewhere in the design I mixed up some drive directions and as a result, wheels on the same drive end would spin in opposite direction. Quite the oversight… In order to fix the rotation direction, we initially tried wrapping chain around and idler and using what would be that back side of the chain to drive the vertical drive shaft. This was acceptable until high loads caused chain skip and eventually master links to explode. The final fix involved adding a set of gears to reverse the direction, there was barely enough clearance to do so but it was done.
Another problem which came up with arranging the Andymark Supershifter gearboxes vertically was gears falling out of place. The shifting mechanism uses a dog clutch and gears with press fit bearings, and the shafts would normally be parallel to the ground in a “tank” drive configuration. Unfortunately, the gears we received from Andymark were barely a slip fit and would fall off their bearings, and we would lose drive power. Some loctite on the gear and bearing fixed that problem.
On earlier swerve drive trains, steering too slowly was a concern, so let’s put in a more powerful motor with less gear reduction right? Turns out we were on the edge of having enough torque to turn the wheels and many Banebots RS-550 motors burned out, releasing their magic smoke. These motors had a free speed of ~15000rpm and went through a 25:1 planetary gearbox and were intended to direct drive an intermediate steering shaft on one end of a drive tube. Since that didn’t work, an additional 5:1 chain reduction was added. Problem solved, the drive modules still rotated quickly, without consuming motors.
The original setup:
Though not as successful as I had hoped, I learned a great amount designing and working on a very complex machine.
The models for the full robot can be downloaded here, and there are more photos below. Keep in mind all this was done years ago when I was still in high school.
Each drive “tube” featured a total of 5 chains and too many sprockets. The sprockets wedged between chains are being used for tension.
2 separate conveyor systems mounted perpendicularly for 3 sided pickup:
Eating some carpet: