Generating Branded Product Concepts: Comparing the Bees Algorithm and an Evolutionary Algorithm

The task to generate product design concepts to maintain a particular brand identity whilst meeting functional requirements is challenging to designers. Shape grammars have been shown to be able formally to describe the creation of branded product shapes using a set of shape rules. These shape rules are applied manually to generate a family of new design concepts that maintain the brand identity of the product. However, shape grammars are not meant to evaluate whether the generated product concepts can meet specified functional requirements. When a shape grammar is combined with an optimisation technique such as an evolutionary algorithm, a computational procedure for generating branded design concepts that can meet functional requirements is established. This procedure evolves the set of shape rules whilst evaluating how well the outcome of the new design concepts meet a specified functional requirement. A new optimisation technique, the Bees Algorithm, was reported to perform better than many existing techniques. This paper compares the Bees Algorithm with an evolutionary algorithm when they are combined with a shape grammar in generating concepts for a new Coca-Cola bottle design. The Bees Algorithm was found to outperform the evolutionary algorithm in this investigation.