We are planning on using the swerve modules we developed in the offseason last year. The design is a bevel-beside-wheel inspired by the SDS modules. We are currently working on manufacturing the parts for the modules. The decision to use swerve was somewhat difficult because it limits us to only 7 motors for the rest of the robot mechanisms, but we feel it will give us a competitive advantage that outweighs the challenge.
To be able to test as many variables as possible, we designed a shooter prototype that can easily be modified to find the best configuration for our goals. From our testing, we found that we will need a dual-angle hood in order to be able to score both from against the HUB and from far away. With this setup, we were able to score reliably in a target less than half the diameter of the HUB.
We prototyped two types of intakes for this year: over-the-bumper and through-the-bumper. The over-the-bumper intake uses a pair of horizontal rollers with a combination of mecanum and compliant wheels to grab, lift, and center the ball. The through-the-bumper intake has a single horizontal roller to grab the ball, and then four vertical rollers to center it before feeding it through a gap in the frame. We tested a wide combination of configurations for both, until both were successful in intaking and centering the cargo, and both could handle intaking multiple cargo at once. In the end, we chose to go with the over-the-bumper option because it uses one less motor and is a slightly simpler design.
We prototyped a mechanism for transporting the cargo through the robot from the intake to the shooter, which we call the cargo conveyor. After trying a number of configurations, we found that the cargo were held securely by a timing belt on one side and rubber stripping on the other. This allows us to run the conveyor continuously with a block in the path to hold the cargo in place, and remove the block to feed the cargo into the shooter.
After considering a number of options, we decided to focus our efforts on a climber similar to the one Snow Problem developed for their Ri3D. After simulating the climbing process and building a small-scale proof-of-concept, we designed a medium-scale prototype to test the mechanism. Unlike Snow Problem, we are powering our arm using a chain and sprocket and using a latch to ensure good grip on the rung.
So far we have been simulating our code using WPILib simulator to check how the subsystems behave, and detect most of the problems before we touch the robot.
Furthermore, we've been calibrating our vision module to detect cargo and the reflective tape on the upper hub. After trying our code, we saw some impressive results, we detected some errors and tested the logic of out taking the opposite alliance's cargo. We chose the sensors we want to use to detect the cargo - REV color sensor V3 and beam breaker. Additionally, we tested PhotonVision on our raspberry pi 3/4 and chose to use raspberry pi 3.