To main content

Real time fusion of sensor data for dynamic 3D mapping

The Norwegian oil and gas company Statoil has developed a new concept for a remotely operated oil & gas platform located offshore. Robots operating the platform requires multiple sensors to provide the information required for operation planning and execution, and fusing sensor data from different sensors is essential to ensure robust and reliable recognition of environmental shapes, features and properties. To utilize this information, carefully tuned signal processing algorithms are required for combining sensor data. A robust sensor system also requires understanding of the sensors characteristics within when operating in challenging environments, and compensation for any undesired effects.

In order to support research on robotic and instrumentation systems for this platform concept, Statoil has financed a robotic lab facility in Trondheim where operations can be simulated in a controlled environment.

Robot contact operations require a close and accurate interaction with the process structure and objects in the lab. Due to small changes in the geometry of the process structure during operation (due to the weight of filled versus empty water tanks), small perturbations in the position of each component in the process structure may be observed. In many cases this may be counteracted by allowing some play between the tool and the structure when designing the tool, but for other operations the accuracy of the robots ability to position itself relative to the process structure is of great importance.

One major activity in the ongoing project Robotics for oil & gas platforms will therefore be to enable a more accurate relative positioning of the robot based on detecting fixed features on the process structure through image analysis. A common strategy to achieve high precision in robot and target localization is to include dedicated markers into the scene. SINTEF has extensive experience in robust recognition of barcodes for images but further research is required to ensure that marker solutions are robust with respect to changes in the environment. Markers will not provide complete target localization, and additional algorithms for high-precision 3D positioning are required. Currently, localization is done either by feature-based techniques, dedicated landmarks or feature detection techniques, or by using Iterative Closest Points (ICP). These methods are either application-specific, have a small convergence basin or do not have the necessary real-time characteristics, and further research is required to ensure real-time localization and tracking during operation.

To accurately locate the robot relative to its environment will be even more important for the real offshore platform due to variations in the geometry of the structure caused by temperature variations and environmental forces from wind and waves. This activity will also be further developed in order to find methods for measuring and mapping permanent changes in the process structure that can be used to update the 3D model of the structure.

 

A robot operating an offshore oil and gas platform requires multiple sensors to provide the information required for operation planning and execution, and fusing sensor data from different sensors is essential to ensure robust and reliable recognition of environmental shapes, features and properties.

Key Factors

Project duration

01/01/2011 - 01/01/2011