Cognite works with customers to leverage robots, drones, and ROVs to efficiently capture inspection data. Data from robot sensors (video, audio, LiDAR data, gas detection, thermal images) are captured and stored in Cognite Data Fusion (CDF).
Avoid sending people into hazardous areas, such as inside cargo and ballast tanks. Replace time-consuming manual routine work to reduce OPEX costs. Automate data capture by letting robots perform regular data collection missions. Increased frequency of data capture contributes to reduced operating risk.
An EX Robotics UGV is used for automated capture of various inspection data. The robot can be controlled manually or perform automated missions to regularly collect data.
Increased volume and quality of visual data enables the use of computer vision. Deep learning models are trained to detect maintenance needs, including surface properties such as rust and cracks, or lighting fixtures that have stopped working. Comparison of data over time can help identify developing issues and misplaced objects. Automated monitoring based on computer vision improves the quality of the inspections and free up time for engineers.
Rust detected in drone footage inside one of the FPSO tanks. Computer vision models assist in finding relevant information to improve inspection quality and accelerate execution.
Cognite takes the raw 3D and video data, and performs optimization and processing to enable advanced visualizations. 3D CAD models are analyzed and transformed for efficient and loss-less rendering of very complex assets. Image, video, and LiDAR data can be processed to generate 3D textured photogrammetry models. The 3D data is contextualized to link all elements to relevant assets, geographic location, and time.
A small section of a 1TB point cloud covering a deck of an FPSO. The point cloud provides an accurate current state of the FPSO deck, and can be combined with regular robot footage for continuous remote inspection.
All asset data can be combined into a complete 3D asset and field visualization, incorporating 3D CAD models, 3D terrain models, scanned point cloud data, photos and videos, 360 photos, and 3D photogrammetry models. The digital representation of the asset can then be overlayed with real-time sensor and event data, as well as providing access to all relevant documentation.
A complete digital twin of an FPSO, combining CAD, point cloud scans, robot footage, and real-time data. The twin provides a single entry point to all relevant inspection data, and avoids the need to access siloed vendor specific applications where data is visualized without context.
Optimized workflows allow users to effectively interact with visual inspection data, recording relevant findings and conclusions. Combine visual analysis with numerical analysis to validate decisions. Integration towards relevant backend systems allow for end-to-end completion of inspections workflows.
The digital twin of the asset can be used for remote data capture management by providing robots with data capture tasks. Photorealistic visualizations enable fully remote inspections. More asset maintenance activities can be planned remotely, reducing the cost associated with service vendors traveling to and from the assets. VR allows for immersive activity planning.