Ingmar’s research focuses on the application of machine learning techniques to emerging mobile robotics tasks such as semantic mapping, active exploration and life-long learning. Mobile robotics presents an exciting and unconventional domain for machine learning applications since the data typically gathered by a mobile robot differ significantly from that of other, more typical application areas: they often originate from a combination of different modalities sensing spatially and temporarily contiguous workspaces. Expert labelling of a limited amount of new data can be obtained by way of human-machine interaction. New data can be acquired on demand. The requirements imposed on machine learning methods in mobile robotics are similarly particular: methods are desirable which are able to process and, if necessary, assimilate new data online and in real-time. The algorithms should be able to exploit the sequential nature of the data and be able to provide accurate measures of confidence to enable robust action-selection. Of particular concern in current work is the extraction of ‘higher-order’ semantic information from sensor data for autonomous navigation and mapping tasks outdoors.