Sensorless pedaling torque estimation by front and rear wheels independently driven power assist bicycle

Abstract

It is important to decrease cost of power assist bicycles for their wide spread. Because force sensors for pedaling torque estimation are expensive, torque sensorless pedaling torque estimation is a good way to reduce cost. Therefore, in this paper, a pedaling torque estimation method for power assist bicycles is proposed. It can be implemented only in front and rear wheel independently driven type power assist bicycles, which were also proposed by the authors. In the proposed method, front and rear wheels are controlled to be the same velocity. Owing to this, a friction coefficient between a front wheel and the ground is equal to a friction coefficient between a rear wheel and the ground. Hence, we can eliminate one unknown parameter, a friction coefficient, from motion equations. In addition, because the proposed power assist bicycles have two motors, both external torques of front and rear wheels can be estimated. Together with data from acceleration sensor and the two estimated external torques, pedaling torque, normal forces on each wheel, and running resistance, can be estimated in real-time by using an iterative least square method. The validity of the proposed method is verified by simulations and experiments.