Safe autonomous vehicle steering is one of the most difficult aspects of an artificial intelligence system. Therefore, researchers have concentrated on training deep network models from front-facing camera data synchronized with the steering angles. In this study, three different end-to-end deep learning models are developed and the success of these models on the Racecar mini autonomous vehicle is evaluated. Experimental analyzes show that proposed models give successful results in steering the vehicle autonomously. The proposed models can extensible so that information such as angle, speed, and brake can be acquired instantaneously from the road view taken from the autonomous vehicle.