For the last week, Team 4904 has been furiously designing, prototyping, and testing the hardware, software, and electronics of our 2017 robot.
The week started with kickoff, where nearly 75 students from Team 4904 and Team 5940 gathered to watch the reveal of FIRST Steamworks. We were immediately excited by the game’s eye-catching field elements and the addition, this year, of human “pilots” residing within them. We began a strategy breakdown, mathematically weighing the advantages and disadvantages of different options of scoring (Fuel, Gears, and rope climbing). We eventually decided to make the Gears our first priority, as we suspected they would be worth many more points per back-and-forth “cycle” of our robot.
Having a rough idea of our priorities, Design and Fabrication teams then began brainstorming, generating three to four possible mechanisms for each scoring method. The group broke into sub-groups with individual members selecting ideas they wanted to prototype. For instance, a subgroup focused on climbing, testing out ways of grabbing ropes with different attachments on the most reliable way. Another sub-group wanted to investigate the repeatability of shooting fuel into the high goal, adapting our 2016 robot’s shooter for the differently sized balls. Additional teams emerged, working on intaking fuel from ground level and on storing and depositing it into the low goal at high volume.
Meanwhile, the Programming subteam also split into specific projects. Subteam members worked on varying tasks, ranging from developing a computer vision pipeline for alignment with the gear lifts to prototyping of autonomous code and updating robot firmware. A few members of the sub team continued their work on LiDAR (Light Detection and Ranging), using a low-cost unit obtained from a robotic vacuum cleaner. The Electronics subteam refined their motor control boards for upcoming prototypes, and finished some offseason training projects (like an LED strip display for our pit). Members also created PCBs—Printed Circuit Boards—to increase the reliability of the CAN network.
During the following few days, many of the juniors (who had been gone on school trips) returned, and were updated on the strategy decisions made during kickoff. Prototypes resumed. Floor intake of balls with grippy, flexible polycord proved to be successful, with one of the season’s first functional prototypes. A slotted gear holder also quickly came together, showing promising effectiveness as a proof of concept. Climber, shooter, and hopper prototypes picked up the pace, working on increasingly more robust and refined systems.
In those few days, the Programming subteam also made valuable breakthroughs. Thanks to a few dedicated members, updated software, and a field-accurate rig, the computer vision system was able to detect and localize the gear deposition peg, making automatic alignment a high probability for the 2017 robot. A few days later, the camera was mounted on our testing drivetrain, and the drivetrain was able to turn to face the peg.
LiDAR work also progressed well, as refinements in communications protocols allowed reading of the sensor’s data to become easier and faster. In addition, the Electronics subteam worked hard in a variety of areas. Electronics members were able to power up mechanical prototypes for proofs of concept. The subteam also progressed in their knowledge of KiCAD, the software used to design their CAN PCBs.
Fabrication during Week 1 wasn’t limited to prototyping. A few days of midnight-oil CAD had produced a full model for an 8 wheel drive chassis that would then be CNC machined in-house. Careful workflow and effective offseason training meant that our fabrication subteam was able to not only build most of the primary frame by the end of Week 1, but also solidify our goal of producing two robots.
At the end of our first week of build season, we had final prototypes. These prototypes tested the assumptions we had about our designs and reflected challenging but rewarding iteration cycles. The climber system ended up with two working designs: one that slotted into a knot in the rope, and one that used the hook side of Velcro to grab the rope. The gear slot went through a number of different iterations, eventually resulting in a laser-cut, piston-actuated prototype that let go of the gear with two arms, reaching around the game piece from above. Our shooter prototype, with laser-cut and precisely assembled ribs, was able to fire two to three balls in rapid succession before losing too many rotations per minute on its flywheel. Due to a quick side view layout drawing in Solidworks, the team finally had an idea of where each system would fit and how the intake, hopper, and shooter would interface with each other.