Most of the existing mobility models use simple, linear motion to describe node movement, and there are few studies on curve movement patterns based on multiple obstacles in real scenes. How to describe the avoidance and curve movement patterns of nodes in multivariate obstacle scenes has become a key difficulty in mobility modeling. To address this, this paper proposes a mobility model based on Hermite curves, which fills the gap in curve mobility models for complex, multivariate obstacle scenes in mobility modeling. Firstly, the paper analyzes the demands of mobile patterns in real scenes, and then theoretically models the Hermite curve direction, combining node movement with Hermite curves to achieve curve movement, and adding perception parameters to avoid obstacles. To study the model's performance in practical scenes, this paper researches the spatial probability distribution of nodes and the spatio-temporal dynamic characteristics of network topology under the model's action, and defines the evaluation indicators of the model. To demonstrate the model's application performance, it is compared with the curve mobility model SRMM and the classic RWPM model. The results show that the model not only adapts to the obstacle avoidance application scenes with multiple obstacles, but also has a more uniform spatial probability distribution of nodes, better spatial coverage, and better spatio-temporal dynamic characteristics of the network topology composed of nodes than the SRMM and RWPM models. This model can provide a more realistic application scenes for the simulation of MANET.