Many people who have to rely on powered wheelchairs find it hard to fulfill daily navigation tasks with their chairs. The SmartWheeler project aims at developing an intelligent wheelchair that minimizes the physical and cognitive load required in steering it. In this paper we briefly outline the SmartWheeler project and its goals. We then argue that it is important to have a standardized test to evaluate intelligent wheelchairs in terms of performance and safety. No such test exists as yet for intelligent wheelchairs, but there has been an effort in the clinical community to design tests for conventional wheelchair usage. We discuss the existing Wheelchair Skills Test (WST). We then suggest a paradigm that allows us to use this test to benchmark the quality of intelligent wheelchairs, and in particular their interface, in a task context that is relevant to clinical practice in rehabilitation.