Assessing assumptions in kinematic hand models: A review

The incredible complexity of the human hand makes accurate modeling difficult. When implementing a kinematic hand model, many simplifications are made, either to provide simpler analytical solutions, to ease implementation, or to speed up computation for real time applications. However, it is important to understand the trade-offs that certain simplifications entail - the kinematic structure chosen can have important implications for the final model accuracy. This paper provides a brief overview of the biomechanics of the human hand, followed by an in-depth review of kinematic models presented in the literature. This review discusses some simplifications that may often be inappropriate, such as assuming no metacarpal bone motion or assuming orthogonal, intersecting thumb axes. This discussion should help researchers select appropriate kinematic models for applications including anthropomorphic hand design, humancomputer interaction, surgery, rehabilitation, and ergonomics. Some modeling issues remain unclear in the current literature. Future work could compare thumb MCP models and better investigate unactuated compliant degrees of freedom in the hand.