Combinatorial layout design has been a challenging task in engineering design as well as operation research. In general, combinatorial layout design is a process of allocating a set of space elements on a layout base, performing grouping and designing topological and geometrical relationships between them according to certain design objectives and constraints. Therefore, a wide variety of application areas ranging from mould layout design, product family design, architectural floor plan layout design and component space layout design to circuit board layout design, web page layout design, nesting design and so forth involve combinatorial layout design. The nature of combinatorial layout design is complex, combinatorial, non-repetitive, generative and human-dependent. Some specialised software tools can support designers to produce a geometrical layout design. However, combinatorial layout design involves a large number of combinations of various layout design alternatives and design considerations. When human designers deal with large and complex layout design problems, they easily get bored, distracted and tend to make mistakes. A good layout design still highly depends on a human designer's experience, knowledge, and creative ability. In practice, it is virtually impossible for human designers to try out all possible design alternatives manually to find the best trade-off solution between cost and performance.
Limitations of traditional approaches for design automation & optimisation
Traditional design automation approaches, such as rule based reasoning, case-based reasoning and parametric design template, cannot produce truly creative, unpredictable or novel layout design solutions because they are unable to imitate human creativity based on pre-processed human problem-solving knowledge or human-generated solutions. Moreover, a solution space of a combinatorial layout design problem is so large that design knowledge cannot be captured, formulated, reused and represented in the form of rules, cases or design templates efficiently. On the other hand, traditional optimisation techniques, such as linear programming, branch and bound and gradient-based algorithms have been adopted to find the optimum strip packing layout design and container stuffing. These traditional optimisation techniques are efficient to search for the nearest local optimal solution with respect to the given initial solution. However, they are limited to a narrow class of simple layout problems where explicit mathematical equations describing the objective functions and constraints are available. In practice, finding a global optimum layout solution to a combinatorial layout design problem cannot be treated as an ordinary design parameter optimisation problem with a fixed number of variables based on a given initial layout design. In addition, layout design objectives and constraints and interactions among them are difficult to build as true mathematical models. More importantly, a search space of a combinatorial layout design problem is so large that optimisation should aim at searching for a population of good layout design solutions rather than a single local optimum one. In addition to the aforementioned optimisation techniques, Heuristic Rule-based (HR) algorithms are commonly used to solve specific types of packing and cutting stock problems. Previous research demonstrated that acceptable layout solutions could be generated efficiently based on special heuristic rules derived from common sense or experiences. However, these HR approaches are only applicable to a specific class of component space layout design problems where well-formed heuristic rules are available. Moreover, because of these reasons, such traditional optimisation techniques are unable to navigate such large search spaces to find near optimum solutions globally and are likely to be inferior local optima.
Our innovative solution
NiCADa Research & Development Ltd. innovates this traditional human-dependent design method and fully automate and optimise combinatorial layout design by combining our Patent-Pending nature-inspired Evolutionary Design technology with advanced 3D CAD systems.
Nature-Inspired Computing Applications for Design Automation and Optimisation