Designing an API at an appropriate abstraction level for programming social robot applications

Whilst robots are increasingly being deployed as social agents, it is still difficult to program them to interact socially. To create usable tools for programming these robots, tool developers need to know what abstraction levels are appropriate for programming social robot applications. We explore this through the iterative design and evaluation of an API for programming social robots. The results show that high level primitives, with a close mapping to social interaction, are suitable for programming social robot applications. However, the abstraction level should not be so high that it takes away too much control from programmers. This has the potential to enable programmers to produce high quality social robot applications with less programming effort. HighlightsWe seek abstraction levels appropriate for programming social robot applications.High level primitives with a close mapping to social interaction are appropriate.But the abstraction level should not be so high that it takes away too much control.Benefits: a close mapping, high abstraction level and high local visibility.Negative effect: low remote visibility, making progressive evaluation harder.