Let software do all the hard work
Software and payload computers to be used on AUVs as well as for post-processing of data
EIVA offers several unique capabilities focused on autonomous underwater vehicle (AUV) operations – both software and hardware – as a result of our vision to provide technology for fully autonomous subsea inspection operations and processing.
One challenge with AUV-based operations is that typically there is very little communication available – this means that a lot of data needs to be either processed onboard the AUV during the mission or after recovery of the AUV.
EIVA offers the fastest data processing solution for AUV operations available, combining both high-speed vessel-based processing with onboard AUV processing.
Our machine learning development team focuses on applying deep learning and computer vision/VSLAM to our industry, and we offer automated object detection, camera-based navigation, pipe tracking and much more based on latest technologies.
- High-speed data processing, where our Workflow Manager software offers unmatched processing times after recovery of the AUV. For example, 50 hours of data can be processed in less than 2 hours.
- Onboard processing, where the Workflow Manager software and specialised versions of NaviEdit and NaviModel without a user interface can be run on the AUV while sailing, thereby reducing the post-processing after recovery by 30% or more. Onboard processing includes sonar data import, cleaning, QC checks for density/gaps/etc, mosaic/GeoTIFF generation and much more.
- Camera-based pipe tracker, where our NaviSuite Deep Learning software can run on the AUV, connected to the AUV camera and in real time detect the pipe, provide video QC, and provide top of pipe to the AUV.
- Automated video eventing, where our NaviSuite Deep Learning software can automatically perform event registration of object types such as pipe, field joint, anode, damage, debris etc – using the camera in real time. This can be used to redirect the AUV to gain more data on points of interest, or simply to save processing time.
- VSLAM scanning using camera¸ where our near-real-time VSLAM algorithm can be used to build dense colour point clouds and 3D mesh models of what the AUV sees – just using a single camera. (To be released.)
A small Microsoft Windows-based payload computer is available for onboard processing and another Linux/Nvidia based computer is available for the Deep Learning software and VSLAM.