An optimal control solution to the predictive dynamics of cycling

Purpose: Pure predictive dynamics aims at predicting the set of driving inputs in the absence of any a priori data and can be applied in movement science to generate biomechanical variables in many different what-if scenarios. The objective of this research was to solve the problem of the predictive dynamics of sub-maximal cycling by means of an optimal control computational algorithm that makes use of an indirect method. Methods: To this, a 2D two-legged seven bodies three degrees of freedom model of the lower limbs of a cyclist has been developed and validated against the average behaviour of eight well-trained cyclists pedalling at different sub-maximal intensities (100, 220, 300 W) at constant cadence (90 rpm). Experimental data adopted in model validation consists of the hip, knee, ankle joint centre and crank kinematics and the right/left crank torques. Results: It has been found that the model can replicate the major features of pedalling biomechanics and the ability of a cyclist to deliver a larger torque if a larger power output is required and the cadence is kept constant. The reported mismatches with experimental data get smaller as the power output increases. Conclusions: It is suggested that: (1) an optimal control based on an indirect method approach can provide a solution to the predictive dynamics of sub-maximal cycling, (2) predictive dynamics adapts accordingly to real data for changes in power output.