A mixture of modular structures to describe human motor planning level: A new perspective based on motor decomposition

A modular hierarchical structure is developed to describe human movement planning level. The modular feature of the proposed model enables it to generalize planning a task. The movements are planned based on decomposing a task into its corresponding subtasks (motion phases). There is a module responsible for one condition. The final plan is constructed using soft computing of the plans proposed by different modules. Each module estimates the kinematics of the joints at the end of each subtask; we call them kinematic estimator modules (KEMs). A timing module estimates the duration of motion and a gating module determines the responsibility of each KEM under different conditions. To evaluate the performance of the proposed model, a set of experimental data are collected for the Sit-to-Stand movement under five different bases of support as different conditions. The results showed that this architecture is able to plan the motion under new untested conditions. Good match between the simulation results and experimental movement, low computational requirements as well as behavioral and neurophysiological supports, make it to be considered as a good candidate to interpret computationally the function of the motor planning level in the central nervous system.