The synchronous control of yaw motion and tilting motion is an important problem related to the lateral stability and energy consumption of narrow tilting vehicles. This paper proposes a method for the tilting control of narrow tilting vehicles: tilting feedforward synchronous control. This method utilizes a proposed novel prediction method for yaw rate based on a recurrent neural network. Meanwhile, considering that classical recurrent neural networks can only predict yaw rate at a given time, and that yaw rate prediction generally needs to analyze a large amount of computer vision data, in this paper, the yaw rate is represented by a polynomial operation to predict the continuous yaw rate in the time domain; this prediction is realized using only the driving data of the vehicle itself and does not include the data generated by computer vision. A prototype experiment is provided in this work to prove the advantages and feasibility of the proposed tilting feedforward synchronous control method for narrow tilting vehicles. The proposed tilting feedforward synchronous control method can ensure the synchronous response of the yaw motion and the tilting motion of narrow tilting vehicles.

Active tilting vehicles tilt to the inside of the corner when the vehicle is steering. The tilting motion improves the steering and roll stability of the vehicle. The steering mechanism and the tilting mechanism of the vehicle are connected in parallel. The transmission of the steering mechanism is influenced by the movements of the tilting mechanism. In order to solve this problem, a parallel mechanism is proposed in this paper. It consists of a spatial steering mechanism and a tilting mechanism in parallel. A mathematical model of the parallel mechanism with the wheel alignment parameters has been established. The model calculates the decoupling conditions of the parallel mechanism. In this study, a decoupling method for the parallel mechanism is proposed. A prototype of the parallel mechanism was designed according to the proposed method. The prototype was found to reduce the influence of vehicle tilting on the outer and inner wheel steering angles by up to 0.64% and 0.78%, respectively. The steering geometry correction rate of the prototype is between 1.198 and 0.961. The correctness of the model was verified by experimentation on the prototype. The proposed method can effectively decouple the tilting motion and steering motion of the vehicle and make the wheels on both sides satisfy the Ackerman steering condition.