While robots have been popular as a tool for STEM teaching, the use of robots in other learning scenarios is novel. The field of HRI has started to report on how to make effective robots usable in educational contexts. However, many challenges remain. For instance, which interaction strategies aid learning, and which hamper learning? How can we deal with the current technical limitations of robots? Answering these and other questions requires a multidisciplinary effort, including contributions from pedagogy, developmental psychology, (computational) linguistics, artificial intelligence and HRI, among others. This abstract provides a brief overview of the current state-of-the-art in social robots designed for learning and describes the aims of the Robots for Learning (R4L) workshop in bringing together a multidisciplinary audience for furthering the development of market-ready educational robots.