Project Overview

The development and use of autonomous vehicles are on the rise and that creates new challenges to solve. The scenario for this project is that a distressed underwater vehicle has been lost during operation and needs to be retrieved. Using information from last years project, a search area with a 25m radius is designated and a BlueROV2, a UUV from Blue Robotics, is used to solve the problem.

Approach

Using the Blue Robotics Ping360 Scanning Imaging Sonar enabled us to estimate a heading and a range to the target through signal processing. By controlling the UUV with a task planner we are able to travel semi-autonomously to the target.

Hardware Setup

Simulation

The simulation environment is built around Bluesim, which is a pre-built simulation environment from Blue Robotics. The environment runs as Software-in-the-loop (SITL) using Godot 3.5 which is a free, open source game engine, and docker container running ArduSub for the controller. Running SITL allows for using the exact same algorithms in the simulation as in the physical UUV.

Signal Processing

The intensities retrieved by the sonar are getting filtered with a certain threshold, which makes it easier to detect important objects in the surrounding area.

After the filtering of intensities is done, a DBSCAN is performed on the data. The DBSCAN function is a clustering method which recognises clusters. The different clusters detected are compared based on mean intensity and size of each cluster. The clusters, shown in the figure above, are represented in different colours. When the target has been identified, a relative heading and distance from the UUV to the target is returned and used within the task planner in execution of the mission.

Task Planner

The task planner is engineered to execute a sequence of operations autonomously, guiding the UUV from submergence to target acquisition with the scanning imaging sonar and manual intervention for mission completion. The planner operates as a finite state machine, transitioning between states based on environmental feedback and preset operational parameters.