According to the signals captured by cameras, lidars and radars, the students will understand the relevant physical principles, and learn and design the deep neural network architectures to realize the localization and recognition of objects in front of a self-driving vehicle as well as understanding subject behaviors for safe driving. In this project, the fundamental training courses will be provided first. The programming skills will be addressed as well. Second, the students will survey the literature about self-driving techniques, and then figure out the feasible architectures for implementation. Finally, students come out the final reports associated with their simulation results, analyses, and discussion.