Maki Lab, Institute of Industrial Science @ University of Tokyo

The redesign was driven by hard operational requirements: high sea mobility, resilience to currents and wind shear, portability, and precise positioning for AUV pursuit. On the mechanical side, I designed an elevated protective frame that shields wiring while doubling as a carrying handle, and engineered the "Sword" — an adjustable, fully removable mounting pole for SSBL communication with a mechanical-stopper adjustment collar. The biggest win was hydrodynamic: I designed a custom NACA 0027 foil to encapsulate the pole and SeaTrac unit, eliminating vortex-induced vibration from the Karman vortex street and cutting the static moment on the joint by 81% — from 41 Nm down to 7.8 Nm — while a precision side groove kept cabling routed and let the SeaTrac height adjust in 200mm steps.

Electrically, I integrated new SeaTrac and side-scan sonar payloads into the boat's existing power distribution, used a MAX3232 IC to step RS-232 signals down to the 3.3V logic the Navigator flight controller expects, and pulled raw MAVLink GPS and IMU data directly from the BlueBoat's internal sensors instead of adding external GNSS hardware — cutting wiring complexity and hull-penetration leak risk. To fold the vehicle into the lab's broader infrastructure, I migrated the whole stack onto a custom 192.168.70.x subnet, configured a topside Linux PC as a gateway with kernel IP forwarding and iptables NAT masquerading, and reconfigured BlueOS's MAVLink endpoints to stream telemetry over UDP to QGroundControl and MAVROS simultaneously.

On the software side, working in KAPPA — the lab's ROS2 stack — I built modules for EKF state estimation, ArduPilot node interfacing, and acoustic modem communication, and single-handedly wrote chatora_viz, an RViz-based real-time diagnostic dashboard for pose estimation, coordinate transforms, and mission goals. After water-tank structural testing and ROS2 integration trials, I deployed CHATORA for field testing in Misaki, Japan, where I ran a 140-meter Wi-Fi degradation test comparing ground-level hotspots against an elevated base station — confirming that Fresnel-zone clearance and tree-canopy multipath fading made elevated antennas necessary for reliable long-range AUV tracking.