Pipeline localisation using sonar

[MadsJJ]

Motivation

To make the drone more reliable, we implement a localisation approach using the sonar. That way we add redundancy and reliability using a perception sensor, in addition to the acoustic pinger. The sonar provides significantly longer range than a camera underwater.

Description

Using the 2D sonar image from the FLS, determine whether it contains the pipeline. If the pipeline is displayed, locate the starting point. The starting point of the pipe can be assumed to be endpoint closest to the drone or origo.

Task suggestion

• Create training data for a ML model using the simulator and/or real world data

• Train an ML model to segment the pipe

  • Using YOLO will output a confidence score, making it easier to detect the presence of the pipe.
  • Using a SOTA sonar segmentation model can provide better segmentation and robustness to noise. Detecting the presence of the pipe will be more difficult.

• Using the segmentation mask, locate the pipeline starting point

  • Can use pipe point closest to drone

Deliverables

• A method to determine wether the pipeline is present in the image
• A method to create a segmentation mask on the pipeline the image
• A method to locate the starting point of the segmented pipeline

References

• FLS: https://norbit.com/media/PS-080001-25_WBMS-FLS-Compact_A4.pdf
• SOTA sonar segmentation model with source code available: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10753270&tag=1

Contacts

@MadsJJ