Comparison of Deep Learning Models in Pothole Avoidance for Self-Driving Car

Self-driving car is one of the automotive innovation technologies that uses a computerized system to control a car without human assistance. Big manufactures have been developing this innovation into the fifth level autonomous technology. This study contributes to create a new system as innovation for this self-driving car to avoid road hazards and potholes. This paper reports the result of the conducted experiment on how the self-driving model is able to avoid potholes using end-to-end approach with Convolutional Neural Network (CNN) as a driving simulator called AirSim. Three different CNN models were tested to compare their performance. The result indicates that all the models were able to evade the road hazards.