Deep Learning-Based Decision-Making of Autonomous Vehicles to Predict Accidents.

The autonomous vehicle system ensures safe driving by certainly controlling a vehicle's motion in the lane to reduce the risk of a collision. The proposed architecture is based on convolutional neural network (CNN) deep learning-based with three convolutional blocks. As output, the CNN framework can predict three different risk factor evaluations, i.e., high risk, medium risk, and low risk, used to predict the accident of the AV. The deep learning-based network had a better accuracy rate and computing time for all MTs with a 70% accuracy rate and an 8-ms computing time for risk prediction in autonomous vehicles. The experimental accuracy demonstrates that our trained method has fruitfully accurate results. The self-driving vehicle model successfully predicts the accident risk during driving a vehicle on the road.