Unmanned Driving System Based on DeepLabV3+ Semantic Segmentation

Recently, semantic segmentation technology has developed rapidly and is gradually applied in the field of autonomous driving. In this paper, DeepLabV3+ semantic segmentation is applied to the driverless system to realize the classification of road scene images at pixel level accuracy, so as to improve the accuracy of collected road information. Then, an automatic control strategy based on the Ackerman unmanned vehicle model with cascade PID front wheel steering control and rear wheel differential control is established to ensure that the vehicle can make the correct decision after receiving the traffic information. The cooperation of the above two technologies jointly realizes the automatic driving safely.