Status: AWARD END DATE: 10/31/2020

Background:

ROVs are high-value assets sent to areas difficult to access. Consequently, they are used in the oil and gas industry for tasks which can be dangerous for human personnel, such as rig inspection. Collision avoidance is a paramount concern to protect both subsea assets and the robots themselves. This is necessary, because servicing an ROV stranded subsea would require rescue missions that scale in complexity. In addition, AUV swarms require low-cost, robust methods to avoid agent-agent collisions.

The Robotic Swarm Control Lab and collaborators have designed and tested tri-axial antennas for underwater AUVs and ROVs [1]. Pairs of these antennas could be implemented to rapidly measure relative 6-DOF range and orientation between pairs of AUVs and/or AUVs and underwater assets.

This study will test procedures for collision avoidance using triaxial magnetic induction and computer vision. In this study, we propose to:

• Design TX/RX Control Circuits for Triaxial Antennas
• Generate 3D Models for Transmission Power Between Antenna Pairs
• Validate models through underwater tests at NASA’s Neutral Buoyancy Lab
• Design safe avoidance control laws for ROVs equipped with new sensors

The main goal is to lay the theoretical foundations and validate a hardware prototype for a multi-sensor navigation-aid system that will efficiently and economically provide collision avoidance for multiple robots deployed for high-risk subsea inspection jobs. We will focus on vision control as well as magnetic induction for the short-ranged path-planning and communication required to fulfill oil rig inspection job requirements. The magnetic induction system will exploit range and bearing information, as well as transmit high bandwidth relative localization data when robots are nearby. This will enable sharing of data gathered by alternate sensing modalities. For this proposal, we will share camera data from multiple cameras located in our ROV.