Advances in sensing, communication, and intelligence technologies have promoted the research theme of “ambient intelligence”; that is, making real world environments sensitive and responsive to people’s needs and goals with the help of invisible sensors and actuators embedded in the environment. Ambient intelligence has thrived in many human beneficial application domains including ambient assisted living, traffic control, commercial promotion, and tourism, to name a few. However, the possibility of the widespread deployment of such applications remains unclear. One of the most critical reasons behind this is the uncertainty of perceived sensor data and the associated consequences of triggering unsatisfactory (or life-threatening in emergency situations) actions. The uncertainty is not simply due to technical limitations of physical sensors or environmental interference, but can be introduced by a crude rule system, an over-fitted artificial intelligence model, or even by human users (e.g., by dislodging or physically blocking the sensors). Whatever the cause, ambient intelligence services cannot assume that their data is “clean”, and must therefore adopt some explicit approach to mitigating uncertainty both in input and output.

This workshop aims to bring together representative members of the industrial and scientific communities that have experience in and concern over the impact of uncertainty in real-world ambient intelligence system design. We will discuss state of the art techniques in uncertainty handling and resolution including Fuzzy logics, statistical correlation models, Baysian models, and evidence theories, to name a few. We will lead a discussion on understanding uncertainties, how to make uncertainty more informative to applications, and how to program with uncertainty.

This workshop is open to any topic related to uncertainty issues encountered in ambient intelligence. Examples include but are not limited to:
• Detecting and classifying sensor faults
• Dealing with non-technical uncertainty issues caused by human interactions with sensors in real-world environments over a long period
• Aggregating uncertain data from heterogeneous sources
• Reasoning and classification in the face of uncertainty
• Characterising uncertainty in knowledge representation
• Uncertainty in the sensor web
• Designing and developing uncertainty-resistant applications
• Experiences of what difficulties that uncertainty has brought for existing application designs
• Programming with uncertainty
• Applications or case studies that demonstrate approaches to mitigating uncertainty