Toward a human-centered UML for risk analysis

International audienceSafety is now a major concern in many complex systems such as medical robots. A way to control the complexity of such systems is to manage risk. The first and important step of this activity is risk analysis. During risk analysis, two main studies concerning human factors must be integrated: task analysis and human error analysis. This multidisciplinary analysis often leads to a work sharing between several stakeholders who use their own languages and techniques. This often produces consistency errors and understanding difficulties between them. Hence, this paper proposes to treat the risk analysis on the common expression language UML (Unified Modeling Language) and to handle human factors concepts for task analysis and human error analysis based on the features of this language. The approach is applied to the development of a medical robot for teleechography