The standard SAE J3016 provides a definition for levels of driving automation; however, shared control algorithms that have successfully been applied to vehicle automation, do not reasonably fit into the levels proposed by the standard.
In this paper, the rich literature about levels of automation taxonomies is reviewed regarding the applicability to all automation concepts. Most definitions provide qualitative rankings of different levels of automation for specific applications. However, from an engineering perspective, a quantitative and generic approach would generate great benefit for the design, analysis and evaluation of human-machine systems.
Thus, criteria for a desirable metric of levels of automation are discussed and a continuous and quantitative measure based on the options available to the human is proposed. The application to an advanced driver assistant system example is demonstrated and future ways to leverage the potential of the metric within and beyond automated driving are examined.