After six long weeks, the build season has finally reached its finale. The leader of each respective subteam has completed their parts for the robot and has detailed our strategy and how that strategy drove the decisions behind how we designed and built our robot. After much hard work and dedication, the complete Redbird Robotics robot has been finalized. This is a huge accomplishment, and we are excited to share what we have built together. Below you will find the detailed reveal of our robot, Droideka.
Overall Robot Design
After careful game evaluation, our team decided that our best option would be to shoot high and travel under the trench. We determined early on that being able to hold five power cells and climb was a necessity to keep our team competitive. Originally, we also wanted to be able to balance on the scale as well as complete rotation and position control. As a team, we determined that trying to add these capabilities would hinder our ability to do tasks we deemed more valuable. With this criteria in mind, we designed a robot to quickly cycle that was robust and reliable each match.
Our biggest design challenge was fitting everything inside such a constrained space. To stay under the height constraint of the trench, many of our mechanisms had to have a height alteration functionality. This functionality was prevalent within our shooting mechanism’s articulating hood and our pivoting single-stage elevator.
Our team decided that a simple and robust drivetrain that could withstand strong defense, while still keeping speed in mind, would be optimal for this year’s game. This season as a team, we collectively sought out a drivetrain that utilized a powerful single-speed gearbox. Through testing, our team found that using three Falcons 500s per gearbox on a 10.26:1 reduction could achieve an adjusted speed of ≈ 13ft/s with the powerful defensive push which we were looking for. Moreover, we chose a pneumatic 8 wheel drive because we wanted a robust drivetrain with minimal rocking. This decision reduces variation in power cell shots from our shooting mechanism and provides lots of traction to resist defense.
This year we used more motors compared to previous years, which caused us to focus on centralizing our electronic panel. This allows for easier access to all motors and sensors on both ends of the robot. By utilizing our laser cutter, we were able to manufacture our precision electronic mounting plate. The plate was mounted on top of the indexer with standoffs so we could utilize the space underneath for wire routing. For the first time, we used a manifold to control all of our solenoids. This allowed us to keep our pneumatics clean and easily accessible. Pistons were vital to our robot this year to cut down on extra motors.
While discussing different collection designs, our team prioritized a mechanism that could center the power cells as it collected multiple game pieces at a time. After many prototypes, we decided to go with a piston actuated dual-roller style collection with polybelt on the inside running both horizontally and vertically in order to give a power cell a centering motion. The collection was designed to have a wide roller in the front that spans around 20 inches to make it as easy as possible to collect. A challenge that we encountered was being able to run the entire mechanism utilizing only one motor to limit the current draw. To solve this, we extended the axle on the middle roller and added bevel gears that transfer the motion onto the axle of the pulleys containing the polybelt cord. Overall, the collector was designed to stay within a small given area that does not go past frame perimeter or get in the pathway of the climber.
The focus of this mechanism was to store and transport all 5 power cells from the collector to the shooter. The indexer utilizes 2-inch straight flex wheels to create a “zig-zag” pathway. This path allows us to minimize the amount of space needed to hold all 5 power cells. The rollers of this mechanism are run on belts and pulleys. One of the design challenges of an indexer is to make a design that will not jam. To stop the balls from jamming, our programming team utilized ultrasonic sensors that sense where the power cells are and help evenly space out the balls autonomously. We have one independent roller that feeds the last power cell into the shooter consistently.
The main goal of our shooter was to be able to shoot reliably from various locations on the field. We initially went through designs that included a two vertical flywheel and a hooded single flywheel. Through extensive testing and tuning, our team found that a hooded shooter design was able to shoot with more precision. Our next challenge was to find a way to be able to shoot from a range of places on the field, not just one set position. Integrating our hood with an articulating mechanism proved to be a challenge as we found that a lip on the hood would make our shots inconsistent. The ‘fingers’, as we call them, solved all of our issues by allowing us to have an articulating hood without a lip because the face of the fingers ride flush with one another. Furthermore, our design team chose to go with a motor and sprocket system opposed to a pneumatic system where we could have a wider range of level trajectories, minimizing our inconsistencies.
When brainstorming the lift design, our team’s goal was to create a strong and reliable climber capable of climbing when the bar is level and to do so quickly. After many iterations, our team decided to go with a single-stage elevator that pivots 90 degrees and extends to reach the bar. At the top of the second stage, we developed a carabiner inspired hook which lets the robot drive into the bar, instead of lowering the top stage onto it. This robust design ensures that the robot will not fall off the generator switch once the hook is on. In order to rest above the ground, our team employed a single direction ratcheting winch system to retract the elevator. One of the biggest challenges we encountered while designing the climber is how we were going to integrate the climber in constraints our team sought out. To fix this problem, our design team spent time brainstorming the position and form of the lift, eventually reaching our 8th design, pictured below.
This year the programming team is working with a lot of different types of code: mathematical calculations, automation, and vision processing to ensure our robot’s maximum potential. For the shooter this year, our articulating hood was the result of a lot of math to ensure its accuracy. After creating a function in a calculator, we set it up so that at each distance the robot is from the target, we knew the angle to shoot at. We use our Limelight vision processor to not only align to the target, but we are also using it to actively calculate distance. This data was used to form a line of best fit, which we put into our code to constantly change the angle of the hood while tracking distance to the target. As the targeting mode is activated, the robot is set to align to the target, articulate the hood properly, and allow the ideal angle for the set velocity of the shot.
Our collection and indexer systems are designed in a way which needs precision in order to be effective, so by placing ultrasonic sensors along our indexer we are able to track the positions of the balls as they move through the compartments as well as automate the process, and in turn, make it more efficient for the drivers. We are utilizing PID loops in our code to maintain the speed of our shooter as well as possible and ensures the maximum accuracy in our calculations. Our autonomous period is designed using the built-in 2048 CPR Magnetic Encoder in the Falcon 500s, as well as a gyro and our shooting mechanisms. It is expected to improve as the weeks and gameplay advance throughout the season.