Building a fully AUTONOMOUS submarine

The biggest challenges for an autonomous submarine are three-fold: ① equipping large capacity batteries for long-duration navigation, ② underwater positioning without GPS, and ③ ensuring reliability for navigating distances hundreds of times longer than previous models.

To minimize drag, a cylindrical (tube-shaped) aluminum pipe design was adopted. By keeping the frontal area small even with increased battery capacity, the submarine can maintain its top speed despite slower acceleration.

The steering system uses a minimal motor configuration for energy efficiency: one BlueRobotics T500 thruster (for forward propulsion) and three servo motors (for fin control) to achieve full-axis rotation. The fin design was inspired by shark fins.

As a solution to the underwater GPS problem, a GPS-equipped buoy is connected to the submarine with a short tether. This limits horizontal position error to about 3m, but creates depth restrictions, so it is limited to initial tests.

The autonomous control program has a two-layer structure: Raspberry Pi (high-level commands: calculating target heading and pitch angle) and Pixhawk (low-level control: attitude control). It calculates the angle and distance to the next waypoint from GPS coordinates and controls depth using attitude angles, similar to an airplane.

Initial tests revealed issues such as an excessively large turning radius, inability to maintain depth, and slow speed. These were improved by increasing the size of the fins and limiting the rudder angle to within 30 degrees to prevent stalling.

A major problem occurred when one of the Chinese-made waterproof servos burned out. To avoid a several-week waiting period for a replacement, an emergency design change was implemented in a few hours, replacing the upper fin with a motor (slightly less efficient but functionally acceptable).

In the final test, the submarine successfully navigated autonomously through multiple waypoints while maintaining a target depth of 2m with an accuracy of 1.18m. However, it was found that the drag from the long tether was affecting depth control.

In the next video (Part 2), the team plans to cut the tether and embark on a full-scale long-distance autonomous mission (distance from a city to a peninsula). Future plans include implementing seabed scanning functionality by adding sonar.

The team's previous manually operated drone, the 'CPS 5', was adopted by the Belgian Navy for practical operations such as hull inspections and mine detection. CPSdrone currently offers a CPS 5 building course for beginners.