← All projects
FIRST Tech Challenge · Into the Deep · 2024–25

HawkBot

Our team's first fully custom robot. In one season it went from a kit-based prototype to a CNC-machined carbon fiber and polycarbonate machine with separate intake and scoring slide systems, color-sensing intake, and a nacelle-style drivetrain.

Role
Design Lead · CAD + Build
Team
3734 Iron Giant (7 members)
Tools
Fusion 360, Onshape, CAM, Blender
Result
1st Place Design Award, 2025
Drag to rotate · Scroll or pinch to zoomHawkBot V2 · untextured CAD export
01 · Context

The challenge

Iron Giant is a feeder team: most members are rookies each year. This season most of the team was new to FTC. The game asked robots to collect "samples" from a central submersible. Robots could either drop samples into a high basket or clip them onto a high bar as "specimens," and fast cycles decided matches.

Previous robots were built on REV aluminum extrusion kits. We had three season goals for the design side:

  • Design a custom chassis with CNC-milled carbon fiber and 3D-printed parts
  • Make it fast, low-profile, and built for this specific game, with odometry pods for accurate position tracking
  • Teach rookies the full design process, from CAD and CAM to manufacturing and testing
02 · Iteration

Three robots in one season

VersionDesignWhat we learned
V1Kit chassis, one linear slide on a shoulder joint (250° range), rotating claw intakeQuick to build and good for testing ideas, but the intake was unreliable and cycles were slow
V1.5Drivetrain V2, custom shoulder joint (280°), claw intake V2Much faster and a more reliable arm, but the arm joint was still a bottleneck
V2Custom carbon fiber/polycarbonate chassis, 3-stage vertical slides, 2-stage horizontal slides, beater intake with color sensors, transfer tray, shoulder-mounted clawFastest intake and most reliable scoring, but more complex to code and drive
03 · Chassis

The nacelle drivetrain

As CAD lead, I spent the summer designing a Drivetrain V0 (about 40 hours of CAD) to move the team from extrusion kits to a custom chassis. I refined it through hundreds of messages with our sister team, our mentor, FTC's North Texas program coordinator, and the international FTC community on Discord. It wasn't built that season because we couldn't afford it yet, but it led directly to V2.

Drivetrain V2 puts each pair of mecanum wheels in a carbon fiber and polycarbonate nacelle. The nacelles:

  • contain the drivetrain parts so they can be serviced separately
  • each hold an odometry pod for accurate position tracking
  • raise the robot and shield the wheels so it doesn't accidentally push samples around

The robot uses more than 400 cm² of carbon fiber and polycarbonate panels, and it has more custom parts than off-the-shelf ones. The downsides were cost, long manufacturing time, and less reuse between seasons. Our sponsorship (below) paid for it.

Turntable render
Time-lapse of CNC-milling our carbon fiber plates
04 · Problem solving

What broke, and how we fixed it

Beater intake. We replaced the claw intake with surgical-tubing beaters driven by a 9:1 UltraPlanetary motor. Two REV color sensors let it reject wrong-color samples and keep the right ones, capturing a sample in 0.5 to 3 seconds. Timing belts funnel the sample back into a transfer tray.

ProblemRoot causeFix
Claw brokeRack-and-pinion gears stripped and placement had to be very precisePyramid-pattern gripper that works even when placement is imprecise
Weak gripNot enough contact area or friction on the sampleGrippers shaped to fit the sample, coated with rubber cement
Belt skippingCAM/CNC errors put holes in the wrong placesAdapters that use the existing holes to mount an idler pulley for tension
Servo stallingIntake motor mounted high above the pivot made the intake top-heavyElastic bands from a rear support beam to counterbalance the load

The belt-tension issue taught me a lesson I still use: design in adjustment. Every drivetrain and belt I've designed since has tensioning slots.

05 · Results

Outcome

~200 hrsof CAD on the robot
83 ptsaverage score with the specimen strategy (vs. 43 with samples)
$2,825from Southwest Airlines, which fully funded V2

We compared the two scoring strategies with match data and chose specimen scoring, which almost doubled our average. Southwest's support came with real engineering management lessons: a phase-based budget (chassis first, then subsystems), timelines required for continued funding, and design reviews that pushed us to balance new ideas against what we could actually build.

Next project

Custom go-kart