The DREAM lab was created in September 2012 by Prof. Cédric Pradalier within GeorgiaTech Lorraine, the French campus of the Georgia Institute of Technology, and the CNRS UMI 2958 GT-CNRS. DREAM stands for Data-driven Robotics for Environment Assessment and Monitoring. Our research interest is the development of automated systems to monitor environments over long periods of time. Automation may be expressed in a large spectrum going from automated processing of large dataset to autonomous operation of a mobile robot. We consider mostly outdoor environment from natural ecosystems such as lakes and forests to industrial sites where there is a need for monitoring and inspection. In both cases, we are focusing on environments where change happens over a wide spectrum of different time scales. When monitoring an ecosystem for instance, local changes happen within seconds when animals are moving, over minutes or hours when the weather is concerned and over months when considering seasonal changes. Among this wide spectrum of events, we are interested in sorting out events that are particularly significant either because they are very salient or because they are very representative. In this context, automation provides not only the ability to handle large volume of data but also to improve repeatability and reliability. Additionally, robotics primarily provides mobility and coverage, which translate into the repeatability and regularity of the monitoring task, as well as the precise localisation of the measured signals. To close the loop, we want to advance beyond long-duration autonomy and towards autonomous and adaptive observation. This will require to integrate the monitored real-time data stream into the decision-making process and the action planning, with the ultimate goal of maximising the information gain for any task and any environment.