With the increasing prevalence of intelligent vehicles in daily life, their safety issues have gradually become a focus of public concern. Any vulnerabilities or defects can have severe consequences for the vehicle and its passengers. This paper reviews the research progress in software defect prediction for intelligent vehicles and emphasizes its importance in ensuring the safety of intelligent vehicles. The paper first outlines the complexity of intelligent vehicle software and its impact on safety, and then introduces the background, process and importance of software defect prediction techniques. Next, the paper analyzes the classification of intelligent vehicle software defects and discusses defect prediction methods for different software life cycle phase, including the development phase and the release phase. Finally, the paper focuses on deep-learning based software defect prediction methods for intelligent vehicles, and provides an outlook on the current challenges and development trends.