Model Predictive Control for sensor-based group training optimization in professional cycling

Determining the optimal exercise intensity is a crucial factor in cycling to improve cyclist's performance and avoid overtraining. Novel sensor technologies enable to optimize the training not only for an individual cyclist but also for an entire group. A sensor-based Team Cycling Training System has been developed to optimize the group training in cycling. This system consists of three major parts: a hardware platform with basic sensors for training data acquisition, a wireless ad-hoc network that establishes the communication among multiple bicycles, and a control algorithm for the optimization of the training. This paper focuses on the development of the control algorithm, a Model Predictive Controller (MPC). The MPC uses a cycling performance model to predict the physical work loads of the cyclists according to current road profile, headwind, speed and position of cyclists within the group. Based on the predicted physical exercise loads, the MPC uses a cyclist individualized dynamic heart rate prediction model to determine the physiological load of each cyclist and regulates the group training by advising the cyclists to change the position in group, to adjust the group speed, or to split the group in such a way that each cyclist can meet his training plan as exactly as possible. Training results of a two-cyclist-group indicate that the MPC is an effective aid for the group training in cycling.