Home        Store        Docs        Blog

AUV University Project Build help


(Josh) #1

Hi
I’m currently attempting to build an AUV for a university project. I’m new to this area so some help with a few questions would be great.
Bit of background :The goal of is for autonomous movement around a glass tank that’s roughly1.5m X 4m X 1m. We currently have two ROVs to work with a Bluerov 1 and a smaller custom build with ROV with 4 T100 thrusters.
We also currently have a Raspberry Pi3b or Arduino mega to run it.
We’re currently using the pi for motor control and planning on using it for object detection using Opencv.

So my questions are:

  1. The Pi has problems with PWM control (can only control in 5% jumps that we’ve found) so we want to get a flight controller to streamline control, gyro integration, compass etc. What flight controllers work well with the Pi? I noticed the Pixhawk is on your site but is it hard to interface with the Pi for object detection? Naivo 2 and PXFMini were other options i saw. The size of the tank also means we need very fine motor control.

  2. Does anyone have any ideas for close range detection of the glass walls? We’ve tried ultrasonic sensors but nearly all their ranges are 20cm in air minimum which comes out close to a 1m in water (way too big for tank). Laser detection was another idea we had but are struggling to start with what to use and how?

Any help on these questions would be greatly appreciated


(Patrick José Pereira) #2

Hello Josh,

Right now we are using the pixhawk (running ardusub) to control and stabilize the ROV and a raspberry pi (running our companion) that do the camera stream and the bidirectional middleware communication layer from the pixhawk to the surface computer.
You can read more about it in https://www.ardusub.com/software/components.html.
To communicate and control the pixhawk, you can use pymavlink, this python library will abstract the mavlink protocol for you. The camera can be connected to the raspberry, and it’s possible to use opencv to do access the camera, do the object detection and pymavlink to send high level controls to the pixhawk.