UniqueAI

Important results from the project

The aim of the project was to test Interactive Evolutionary Algorithms (IEA) to solve design problems that are ruled by qualitative requirements or have a complex solution space that cannot easily be captured by quantitative requirements. It is assumed that this type of problem is best solved through human judgment. The aim was to test the IEA on these problems in an industrial environment. This has been done in three practical cases where IEA were used with human judgment in an interactive optimization process. The results show that the IEA can be used to solve this type of problem.

Expected long term effects

The tests conducted in the project show the utility of interactive evolutionary algorithms (IEA) to solve problems that require human judgment to distinguish between what is a good and a bad solution. Examples of this are if something is aesthetically pleasing or if a product has a good haptic feel. Other cases may be when human judgment can find better solutions than computer-based analytical tools, if one considers the investment that must be made in an analytical tool that can exceed a human judgment by an experienced designer.

Approach and implementation

The project has been a collaborative project between RISE, Chalmers and three companies that have contributed with practical cases. Co-ordination has taken place through joint co-located meetings. In the sub projects of the practical cases, different sets of researchers have participated together with each company. The subprojects have worked relatively autonomously in the project with information transfer at the joint meetings. Academic supervision has been provided in the subproject at Kongsberg Maritim, where a newly hired doctoral student from Chalmers was active.