AutoPack - Automatic packaging of pipe ant tube installations based on optimization and machine learning
|Funding from Vinnova||SEK 3 575 944|
|Project duration||November 2017 - October 2020|
|Venture||Maskininlärning - FFI|
|Call||Machine Learning - FFI - 2017-06-13|
Purpose and goal
Digitization of industry allows for increased efficiency of the development process for complex integrated products, such as modern vehicles. These systems generate large amounts of data, both during product usage, but also during the development process. AutoPack develops automated engineering tools based on knowledge extracted from these data-sets using methods from machine learning, design automation and multidisciplinary optimization. AutoPack will automate packing of pipes and hose engine installations in.
Expected results and effects
AutoPack seeks to increase the degree of automation in the product development process within the automation industry, through an automated design process based on knowledge and data are extracted from the development process. AutoPack delivers a decision support tool that automatically optimizes resource intensive packing work for pipe and hose installations within the engine compartment. The goal is to achieve a 50% reduction in the number of engineering hours spent in the development process, and corresponding cost savings.
Planned approach and implementation
The project partners are Linköping University (lead) and Fraunhofer-Chalmers Research centre Industrial mathematics, Volvo Car Corporation, and Intelium Engnieering. The project is divided into four work packages (WPS:s) where WP1 develops models and methods for design automation. The models are then used to find optimal solution within WP2, and machine learning algorithms are being developed in WP3, where also an integrated decision support framework is developed. Finally, the framework is implemented at Volvo in WP 4 in order to validate and verified the results of the project.