A Lot Of Hardware Work These Days

Silje Greidung Head of Marketing

Ellen Iren Johnsen Writer (Members since 2021)

New week, new Sunday workshop! Here is a summary of our tasks this week.

The "pizza board" with its components at the moment.

The electronics group's tasks are progressing quickly. Tina and Hege have completed the last battery pack, and Sander has made the Power Distribution Board (PDB), which will distribute power to the whole drone.

Christian has, as of last Sunday, been attaching components to the “pizza board” which goes in the electric housing. So far, it consists of (from bottom to top) an Inertial Measurement Unit, which measures angular velocity and linear acceleration, two relays to convert small electrical stimuli into larger currents, an Electronic Speed Controller which controls the thrusters, and an ethernet switch (on the backside).

See how much smaller this year's PDB (upper) is than last year's!

The acoustics department have been working on one of the tasks for RoboSub. This task includes, amongst other things, an object that will transmit a given frequency in the water. Our team has worked on making software that will remove other irrelevant frequencies from the data. The code is mostly finished, so we are starting software-tests on the code.

Our perception team has been working on two tasks this week. Firstly, reading through and editing old software that deals with data from the sonar, and secondly, improving how we send data from the camera to the Robot Operating System.

Finn from the Autonomous group worked with target following simulation this Sunday.

Finn simulating the target following.

Three models are present in the simulation: the virtual target, which is the one we try to follow. The virtual target moves discretely, which means that for example a 90 degree turn happens in one time-step. This is not possible to imitate in reality, and therefore we have simulated a model of the drone, which the third model – the actual drone, will follow.

The hard-working Mechanical members Sebastian and Håvard has spent the day in the workshop. One of the differences from Beluga Mk1 is that Beluga Mk2 will have an adjustable camera angle. Since Beluga Mk1 tilts forwards in the water, the camera points more downwards than desired, and it's favorable with a camera which can be adjusted easily, in case the situation is the same for Beluga Mk2.

Håvard (left) and Sebastian placing the drill holes.

So, five holes with 7.5 degrees angle between them have been drilled for the camera adjustment. In addition, the el house brackets have been improved and are ready to be 3D printed again, the torpedo is designed and printed, and the lights and the Gemini Imaging Sonar has been attached to the drone. What is left for now is the buoyancy test and possibly adjustment of this, in addition to putting all the pieces of the AUV together when that time comes.

Moving down to the basement, mechanical leader Truls is by the pool, testing the different torpedoes accuracy and precision to determine which design is the most optimal, regarding shape of the nozzle, tail and the length of the torpedo.

The torpedo-shooting setup.

A handful of torpedoes in different shapes have been 3D printed, and each torpedo is tested around 10 times to document the variance, and make sure the sample size is big enough to conclude with which design is the best. Perception leader Benjaminas has spent hours making a Python program which uses object detection of the videos from the GoPro to determine how far off from the middle the torpedo shoots – and it’s awesome!

Here is an offset of -0.07. This means 0.07 of the long lines, which are 50 mm in width, meaning it has an offset of 3.5 mm left from the middle.

The first few torpedoes have been tested, and the results so far is giving us an idea of which designs are best suited for the torpedo.

Easter and pool testing is approaching rapidly. Things are starting to come together now, although there is still much to be done!