Self-driving cars are a dynamically growing industry, fundamental to transportation and robotics, providing a higher demand for AI specialists. Nevertheless, as occurs with so many things, the theoretical understanding of abstraction in textbooks and modern curriculum is not as easily implemented by students. While numerous third-party companies are available in the market that provide kits and real-life guides and practices for self-driving cars, these products are often only for individuals and have quite high prices, making it difficult to apply in the classroom model. To solve this problem, this paper introduces the low-cost Autocar toolkit that university students can use to learn autonomous vehicle programming, which includes basic computer vision algorithms. Learning materials are designed for project-based learning, the ability to practice in groups, and implementation principles that focus on applying the approaches and techniques in computer vision. This approach is intended to address some of the difficulties the institution faces in procuring educational facilities and remove the gap between theory and practice in autonomous vehicle-related education. The AutoCar kit is suitable for classroom practice with groups of 3 members. This paper describes the hardware kit, the software materials, and the exercises. The result is an educational technology tool that contributes to training programming engineers for self-driving cars.