|Customer: IMAGINiT Technologies|
IMAGINiT Technologies, a division of Rand Worldwide, advances the way architects and engineers in the construction, infrastructure, manufacturing and facilities management industries manage their projects. Many Fortune 500 companies work with IMAGINiT Technologies to get a competitive edge through their technology consulting, implementation, training, and support services. As one of the world’s largest implementers of Autodesk 3D design and engineering software, the team leverages unrivaled industry experience to design systems that accelerate innovation while improving project quality and profitability.
IMAGINiT was using the original version of the cylinder recognition tool in its Scan to BIM software. This tool could only recognize a single cylinder at a time and had very low performance on large point cloud data. This did not allow the tool to be widely used and made the reconstruction of models from 3D scans of plants containing a lot of cylinders, such as petrochemical plants and refineries, especially difficult. Automated methods for the detection and fitting of cylinders in point cloud data are essential for 3D reconstruction of these sites. The current cylinder recognition tool did not perform as well as their competition and IMAGINiT wanted to provide their customers with a tool that would outperform in both accuracy and speed.
AMC Bridge is an expert in handling 3D modeling data and also has an extensive background in developing applications based on the PCL library. The AMC Bridge technical team reviewed the requirements from IMAGINiT and quickly assessed the scope of the project. They proposed a solution that would allow the cylinder recognition tool to recognize multiple cylinders at once while simultaneously improving both the performance and quality. AMC Bridge then provided the client with a detailed development schedule and proposed a development team.
At the beginning of the project a QA Lead Engineer developed a set of test plans to be executed at different stages of the development cycle, to ensure functional completeness according to the specification as well as final product quality.
The project focused on three major tasks:
Software engineers worked to optimize the segmentation algorithm used by the PCL library and was able to improve its efficiency by approximately 4 times. The algorithm was also scaled for parallel processing with a utilization rate of 95% per core.
The point cloud was significantly reduced by applying mathematical, statistical and algorithmic methodologies that helped to normalize the density, remove complexity and improve accuracy in surface recognition.
As part of the post-processing the new tool could reject all cylinders that violated the minimum user constraints based on both - the minimal cylinder depth and the area ratio between inliers and the cylinder. It also helped to refine the quality of the finished cylinder.
The quality of the work delivered by AMC Bridge and the overall efficiency of the development process exceeded the client’s expectations. This was possible due to AMC Bridge’s:
- Consulting with IMAGINiT Technologies staff to identify and address gaps in the original
- Designing the final cylinder recognition tool that surpassed competing solutions in efficiency and accuracy.
- Excellent project communication, which ensured that the customer always had upto-date information about project progress;
- Maintaining communication throughout the project development to ensured that the customer was always up-to-date with the progress of the project.
- Delivering the final product within the scheduled time and under budget.