Research Initiatives

Research Initiatives

Research Ideas

The goal of the Third Network Meeting in Bath in September 2005 was to identify areas of research emerging from the cluster work. The result of this meeting is available
here

Research Proposals

In February 2006, the Call for Papers for the first round of research proposals of the Designing for the 21st Century program was announced (see PFD ). This network submitted three proposals; summaries of the proposals are listed below.

Nature Inspired Approaches to Creative Mass Customization

New manufacturing technologies have the potential to revolutionise the manufacturing industry. Digitally controlled manufacturing processes, such as rapid prototyping (RP) machines and modular assembly, allow manufacturing models where each item produced is customized for an individual user, without introducing significant additional costs. The Internet provides a widely available interface, and a number of manufacturers have recently started offering mass customized products, including such high profile companies as Nike, Lego and Dell.

Currently the customization available to the customer before the product leaves the factory is very limited and often superficial. In one common model in the clothing industry, the user has no real design influence at all – the customization concerns the sizing, and the system automatically adapts the shapes of the items (clothes, shoes). The user has slightly more influence in systems where he can chose between a number of predefined options (e.g. parts of different colour) in a modularized manufacturing process. At best, the user can upload images that are then printed onto the object.

The manufacturing technologies to take customization further and integrate it with the design and manufacturing process itself are now available, and there is a considerable market opportunity. What is missing is a suitable design process, which can allow the user to creatively modify a design but ensure the result still incorporates the functionality and design characteristics of the professional designer’s original. This project will develop a novel, computer supported design process which combines designer and user input into a final artefact; developing new manufacturing models where the customer can be involved together with the designer in a creative design process.

Evolutionary design processes with user selection (are limited in the outputs they can produce. The inclusion of a bio-inspired development process removes this limitation: while the evolution always operates on the same, simple set of data (or genes) the development process can transform this data into highly complex forms. Most importantly, style and other properties of the resulting design are determined or at least strongly influenced by the particular transformation used. With an appropriate transformation, we can ensure that all designs the system produces conform to a style developed by the designer; while functionality, manufacturing and other constraints are also enforced. Style and external constraints can additionally form part of the selection process: the evolutionary system can produce far more potential designs, while only those that the system determines to conform to the constraints are presented to the user.

Morphogenic Design

Design is driven, not just by function, but also by the manufacturing processes that are available. Nature uses very different construction methods from those used for man made objects, and that is why it is generally very easy to distinguish between natural and man-made items. Natural morphogenesis is a highly complex, nonlinear process. There are a number of different mathematical and computational models that attempt to model aspects of morphogenesis at different levels of abstraction – for example Cellular Automata, Reaction Diffusion (R/D) systems, and Lindenmayer systems.

We propose a novel design approach based on natural designs, where the designs are defined in terms of particular reaction-diffusion equations, and artificial evolution is used to derive these equations. Reaction-Diffusion? systems are very powerful models, which have been successfully used to model 2 and 3 dimensional formation of form and patterns. In 2 dimensions, they have been used to model the generation of stripes and other patterns on animal coats, and the formation of fingerprint patterns. In 3 dimensions, R/D based models have been successfully used to show patterns of limb formation equivalent to those in chickens. As a result of the non-linearities involved, it is generally very difficult or impossible to 'reverse-engineer' a particular R/D system from the desired or observed output.

The system we suggest uses artificial evolution to construct reaction-diffusion type systems with particular user-defined properties. The output of these R/D systems can be interpreted in a physical form, and one or more user-defined performance values generated. These values in turn serve as 'fitness' to guide the artificial evolution. The artificial evolution will work on two levels, co-evolving the equations and the parameters of those equations. The final designs can currently be implemented using modern 3d manufacturing, and computer controlled machining. A longer-term aim is to use the design process to design for new manufacturing processes based on real processes that can be described as R/D systems.

If we can use some of the principles of natural morphogenic processes to drive our design process then we have a route to developing naturalistic designs that are significantly different from conventional ones, both in terms of appearance and performance. We can also use the same design processes to construct models of natural structures from a limited amount of data.

Ambient Art

We are becoming increasingly exposed to more and more digital information streams, and need effective ways to monitor these without being overwhelmed and distracted from our real-world experiences and primary goals. This is particularly relevant for long-term interactions with almost invisible technologies, which are part of the 2020 future, which will tell us for example where our children are, how our investments are doing, what the state of the power plant is, and so on.

We cannot spend all our time looking at this, but need to know when things are okay, and when they are not. If this information is to be permanently displayed in our periphery, it has to be aesthetically pleasing – we have to want to have it around, and to enjoy interacting with it. Nature provides the inspiration for the design approach we will follow: it offers many forms that are aesthetically pleasing to us, and has processes such as evolution that produce pleasing, effective solutions. The project proposes to investigate how to provide aesthetic informative representations via three strands:

1. Natural representations (e.g. images from nature, which appeal to us somehow)
2. Nature-inspired representations (e.g. evolutionary art pictures) which develop according to nature-inspired rules
3. Human-created art, inspired by natural scenes or processes

Each which will be used to represent complex, large-scale information without presenting all the details. The project will then evaluate these approaches, looking at both their aesthetic qualities, the increased awareness of the information presented engendered in the observer, the degree of attention required by the observer (i.e. do they have to actively concentrate on the images, or can they simply absorb the information without any conscious attention?). The project also uses the opportunity to investigate the psychological mechanisms of non-attentive displays through its experimental process. This work is the first detailed attempt to understand the mechanisms behind the effective use of non-attentive information displays from both an artistic and a cognitive perspective.

---