I have just been successful in securing funding for a PhD studentship. The project is going to be in the area of ‘Shaping the Future of the Intelligent Home’ and is for three years. The scholarship will be advertised in September for start this autumn in January.
Some notes on the project:
Computing devices are becoming ubiquitous. Even though chip implants are not imminent, a multitude of computing devices are being integrated into many consumer electronic products and home appliances, thus becoming part of everyone’s home. Many of these devices will attempt to assist human beings in various aspects of everyday life. Examples of this kind of functionality include the ‘suggestions’ given by modern digital video recorders, also called personal video recorders (PVRs), such as the Virgin TiVo; location-based alerts given by mobile phones; and advice based on the data provided by Nike trainers and communicated to MP3 players and watches. These developments are rapidly affecting our lives and the places we visit regularly. The home is one of the first areas targeted by numerous companies in the home entertainment and communication sectors. Initial developments are in the areas of entertainment (recommendations and defaults in TV viewing by exploring and learning of habits, semantic search for similarities, lighting preferences), e-commerce (assisted home shopping and automated supply control), and energy-saving (automated dimming of lights and setting of the heating according to habits, combined with energy saving patterns).
The home of the future will use a variety of sensors to perceive and learn about the habits of the person(s) using any of the room(s). Partial information will be gathered and put into a larger context by communicating agents. These agents form a multi-agent system in which agents communicate to exchange information to achieve the desired goal by picking suitable plans (rules) and making them the current intention. Several intentions can be active concurrently in the pursuit of a number of (sub)goals. Agents also might have to work together (form a coalition) to achieve the goal(s). Changes in the environment, such as the availability of sufficient resources, can lead to the necessity of dropping an intention and picking an alternative plan.
The user profile will be updated continuously in order to give informed advice and suggestions. The user(s) will not need to interact with the learning system, but – especially in the initial learning phase – it will be beneficial if the user can rate the suggestions made by the intelligent home to accelerate the learning process. Modern, flexible work patterns impose the necessity to move house more often than previous generations. To accommodate this development, it is essential that the information gathered is separated into location-specific information and more general information that would be portable and might even be carried on mobile devices (or chip implants) to be available to the user at any time and at any location. For the home of the future, this could accelerate the learning process in a new environment, since some location-independent personalised patterns would already be available. Potentially, this information can also be used for personal assistance in mobile devices.
This scenario poses some fundamental research questions:
• How can artificial intelligence in general and agent technology in particular be facilitated successfully for assistive devices in a typical home setting?
• How can habits be perceived using affordable technology?
• How can life-assistance agents in general, and home-assistance agents in particular, be implemented according to ethical and legal requirements?
• How can data protection be guaranteed?
• How can unwanted interference with the various sensors, computing devices, etc. be avoided or be kept to a minimum?
• How can the system react optimally and reliably in the case of insufficient resources or other changes in the environment?
• How can conflicting interests of family members and intentions competing for resources be dealt with?
Facilitating techniques known from artificial intelligence (including games AI), learning theory, and agent programming, the aim of the project is to develop algorithm prototypes for an assisting home of the future. Applicable techniques include resource-based and location-based reasoning, rule-based systems, neural networks, search strategies, and a variety of logic and agent-oriented programming methodologies. These well-explored techniques can be combined to form framework for a ‘personal assistants’ in the home of the future.
