Product family optimization with solution-compensation spaces

The joint optimization of components of a product family and their combinatorial arrangement in product variants can be very expensive. A previously introduced approach keeps numerical cost at a moderate level: common components are chosen from the overlapping regions of so-called solution spaces, i.e. permissible ranges for component properties. Unfortunately, these basic solution spaces are small, and results are far from optimal. To produce larger overlapping regions, so-called solution-compensation spaces are adopted for two different types of variables: A-variables affect product performance, cost and commonality, while B-variables only affect product performance. The approach is applied to a product family of automated vehicles to minimize cost associated with a family of perception sensors that are shared among the vehicles. A- and B-variables correspond to sensor characteristics and positions, respectively. Compared to basic solution spaces, numerical cost increased; however, economic cost of the product family was improved by 39%.