My research activities lie in computational behaviour analytics in wearable, mobile and ubiquitous computing: the use of machine learning techniques, miniature sensors, and online data sources to recognize and understand human behaviours, activities, and context, including social interactions and cognitive-affective states. This science enables systems capable of reacting to user commands for explicit human-computer interaction (HCI). More fundamentally, it enables implicit HCI: systems that understand and automatically react to the user's context and needs intuitively and in real-time. It also allows the objective micro analysis of subtle behaviours in individual users, as well as the macro analysis in the large: over long time scales, or for large sets of users. Applications find themselves in wearable/mobile and ubiquitous computing in the form of activity- and context-aware systems and assistants. They include e.g. HCI, activity-aware entertainment, industry worker assistance, sports trainers, lifestyle and health assistants. Computational behaviour analytics can provide quantitative objective data in fields such as pervasive healthcare, psychology, social sciences, or for smarter energy management. I emphasize novel sensing modalities and embedding this intelligence in miniature and mobile devices, multimodal sensor fusion approaches that scale to sensor-rich environments, and follow a lifelong adaptive machine learning paradigm for use in open-ended environments. The fundamental underlying scientific challenge - and my long-term objective - is to eventually reach human-like action perception. My research approach consists in identifying methodological challenges raised by real-world applications. They guide the investigation of novel algorithms or principles. In turn, these methods support the realization of more advanced applications. Among others, I initiated and coordinated the EU FP7 FET-Open project OPPORTUNITY, where we investigated novel methods for context-awareness in opportunistic sensors configurations. I carried out my PhD at the Laboratory of Intelligent Systems of EPFL, where I graduated in 2005. During my PhD I developed bio-inspired electronic circuits with fault-tolerance, learning, and developmental capabilities that were applied to the control of autonomous mobile robots and to signal processing. This work was carried out in context of the EU FP5 FET project POEtic.