Nowadays, with the development of automotive industry, more and more functions and devices are assembled in cars to improve driving experience. Meanwhile, traffic accidents are increasing in recent years. Operations of in-car device human machine interface (HMI) will cause lack of concentration, which is a major cause of the accidents. Common in-car device HMI systems are based on optics, acoustics, and so on, which are faced with environment limitations such as influenced by illumination conditions. In order to deal with these limitations, in this paper, an infrared array sensor is applied to construct a hand gesture recognition system for in-car devices control. The proposed system can overcome disadvantages of other systems and has a wider application. In the system, seven different shapes of hand and movement toward four directions are combined to achieve the aim of device operations. In data processing, convolutional neural network (CNN) is applied to realize recognition. Simulated experiments are conducted to verify the feasibility of this system.