Autonomous Vehicle Navigation Based on Vision and Mapless Strategies

Vision camera recognition system has many roles as face recognition can be used in the criminal system, object recognition can quickly classify items, and behavioral recognition can be placed in the intelligent cockpit to identify and analyze the driver's behavior and make the correct judgment and decision-making. Vision cameras can also be used in automatic automobile driving because the camera will not be distracted and can be installed with multiple cameras to observe simultaneously, so the camera can constantly observe the surrounding situation, which is incomparable to human vision. Mapless traveling is also probable to happen in life, so it is a very common situation. The use of mapless and camera recognition systems is also extensive. In this paper, I will introduce three kinds of autonomous driving. Firstly, I will introduce optical flow, which is made by stacking images frame by frame and calculating the speed and position of objects through the position of each frame; secondly, vision tracking, which is caused by camera shooting and calculating to localize the other objects and make the right decision for the vehicle; and lastly, the no-map strategy, which also combines the camera and radar, and combines the two. Finally, the mapless strategy can also combine camera and radar, combining the advantages of the two to make the data of autonomous driving more accurate so that autonomous driving will be safer for drivers and passengers.