This week, I successfully got the user detection system fully running on the Raspberry Pi 5 with the Raspberry Pi Camera Module. I resolved the remaining camera and runtime issues and verified that the full vision pipeline works reliably on the device. The system now runs the MoveNet pose estimation model in real time, detects body keypoints, computes the user’s torso center, and outputs the position error relative to the center of the camera frame. I also began working on communication between the user detection script and the motor control subsystem, preparing to use the error signals to drive the pan and tilt motors.
My progress is now on schedule. With the major debugging issues on the Raspberry Pi resolved, the vision subsystem is stable and ready for integration. This puts me in a good position to move forward without impacting the overall project timeline, while the rest of the team continues advancing the mechanical assembly in parallel.
Next week, I plan to complete the integration between the vision system and the motor control subsystem so that the umbrella can automatically adjust based on the detected user position. I will also continue testing the system in real-time conditions to ensure stability and responsiveness, and support full system integration as the mechanical components are finalized.