approaches
self_driving car has two approaches now:
- robotics approach——fuses output from a suite of sensors to directly measure the vehicle surroudings and then navigate accordingly.
- deep learning approach——let self driving car learn how to drive by mimicking human driving behavior.
what’s project you will build?
computer vision, deep learning and sensor fusion——find lane markings
- 1.computer vision tell you where to drive
*2. lane finding, use cameras,learned how to extract things like lane markings and other vehicles
- 3.deep learning ,train a neural network to find traffic signs and images
- 4.behavioral cloning ,copy and clone human behavior such as steering actions and brake and gas actions,learn from some examples,neural networkings and deep learning to camera images and do some of the very things that a self-driving car ought to do right
*5. sensor fusion integrate different types of sensors.like LIDARS radars ,cameras gyposcopes inertial measurement uints
-
6.localization build maps of environment with lots of detailed information,alignment of sensor data ,the current posiiton to a map
-
7.path planing ,find a valid sequence of steps or actions in a maze. a path planning algorithm is an algorithm that finds that sequence of actions for you.
-
8.PID contol ,a computer program that steers your wheels,your gas pedal your brake pedal so as to meet a given objective.
*9. put your code onto a real car.