Project A4 - Crossmodal joint sparse feature learning in robot tasks

PIs: Prof. Dr. Changshui Zhang, Prof. Dr. Jianwei Zhang

The goals of project A4 are (1) to investigate a novel approach to robot perception, which is based on machine learning with crossmodal features, and (2) to apply this new approach to specific perception tasks for service robots, providing elderly assistance with everyday tasks. The work is planned in four stages, including the collection of crossmodal data and features, the study of crossmodal methods most suitable for these features, the design of open-ended classification methods that allow online introduction of new classification categories, and the implementation and testing of the resulting strategy.