Year 1, where we ended up

What worked

The vision and prediction pipeline is the year’s primary achievement. The stereo camera system triangulates birdie position in real time, and an Extended Kalman Filter fused with a physics-based aerodynamics model predicts landing position and time-to-arrival.

Mechanically, the robot is complete, both in design and fabrication, except for the motors. Electrically, the power system is mostly finished and is awaiting motors to begin integration testing. The signal system is still under development.

What did not

Motor lead times prevented full assembly, so the Year 1 goal of catching a flying birdie was not met. The primary constraint of Year 1 was motor lead time, preventing the physical robot from being assembled and tested.

Birdie detection under gym lighting remains unsolved and is the highest-risk item for Year 2. HSV thresholding requires sufficient colour information to work, but the short exposures needed to avoid motion blur at high frame rates produce images too dark for reliable thresholding. There is a narrow operating window where exposure is high enough for detection but short enough to avoid blur, and we were unable to find it reliably. Whether background subtraction can close this gap is unproven.

Robot localization remains an unmanaged risk. Since our stereo vision system is reliable, we decided to pivot to a stereo vision approach, using ArUco markers on the robot. This simplifies the localization setup, as the birdie tracking system and the robot localization system can share the same stereo vision hardware and a common coordinate system.

Next

The end goal of a fully autonomous robot remains achievable, but is contingent on resolving birdie detection and robot localization. The team recommends prioritizing background subtraction as an alternative to HSV detection before any further integration work. Across all three subsystems, the core theory is well-developed and individual components are progressing. The path to a functional system in Year 2 is clear.

Over the next year, rapid development is required for robot motion and control system development in order for the robot to reach the required locations to strike the birdie. One key challenge, along with subsystem optimization, is integrating each subsystem in a manner which fits the time budget for hitting the birdie. Timing and telemetry will be key considerations for Year 2.