robocon-r1-robot

R1 was Team Robomanipal's primary robot for DD Robocon 2025 — a vision-guided manipulator capable of picking, orienting, and placing objects in a defined arena sequence. The robot used an omnidirectional chassis with four mecanum wheels for precise positioning, and a 3-DOF arm with a custom gripper for manipulation tasks.

Robocon 2025's challenge demanded sub-centimetre placement accuracy in a dynamic competition arena — impossible to achieve with pure encoder odometry. Drift accumulates, arena surfaces aren't perfectly flat, and opponents can displace your robot mid-task. The core engineering problem was: how do you localise and act accurately without external positioning infrastructure?

I led the embedded systems and PCB design for R1. Designed and fabricated the main controller PCB (STM32F4-based), all motor driver boards, and the sensor fusion board. Wrote the BLDC FOC control loop in Embedded C, integrated the depth camera (Intel RealSense D435) with ROS 2 running on a companion Raspberry Pi 5, and implemented the inter-board UART communication protocol.

We finished AIR 22 nationally at DD Robocon 2025 — top 10% of all participating teams. Missed ABU Robocon qualification by 10 ranks. The depth camera + vision pipeline achieved reliable localisation within ±3 mm under competition lighting. The custom PCBs had zero field failures across all matches.