Go to the profile of  Suraj Patidar
Suraj Patidar
Follow
I am machine learning and deep learning enthusiast.
2 min read

Self Driving cars are the upcoming future

Self Driving cars are the upcoming future cars and trucks running on the roads.Time is not far enough when you will be watching autonomous cars on the roads.

Self Driving cars are the upcoming future

Self Driving cars are the upcoming future cars and trucks running on the roads.Time is not far enough when you will be watching autonomous cars on the roads.Big Tech Giants are working on these technologies since past many years but not came up with succesful output.SDC journey begins since 1986 when Carnegie Mellon University Navlab built one of the first self driving cars that was ever controlled by a computer then in 1995, Mercedes Benz completed the Eureka Prometheus Project which was the largest autonomous vehicle research and development program in the history up to that time.This program redefined the state of the art for self driving vehicles and then in 2005,Sebastian Thrun who is also the founder of Udacity led the Stanford racing team that won the DARPA Grand challenge, a 100 mile self -driving car race through the California desert.After winning the DARPA challenge Sebastian joined Google and started the Google self driving car project in 2009.


Main approaches for implementing self driving cars platform are:
1)HD Maps
High Definaion Maps underpin almost every other part of the software stack including localization,perception,prediction and planning.


2)Localization
In these it is how the car determines its location in the world.In these the car utilizes laser and radar data and compares what it sees through these sensors to a high defination map.This comparison enables the vehicle to localize itself with single digit centimeter level accuracy.


3)Perception
In these we will look how the self driving car sees the world.Deep learning is an important and powerful tool for perception.CNN's also plays crucial role in perception tasks such as classification,detection and segmentation.These approaches work with data from several different self driving car sensors, including cameras ,radar and lidar.


4)Prediction
In these we will outline ways to predict how other vehicles or pedestrians might move, for these will we take the help of RNN which tracks the objects movement over time and uses this time series data to predict the future.


5)Planning
This will cover how to combine prediction  and routing to generate a trajectory for our vehicle.This is the hardest part of self driving car.


6)Control
This part will teaches us how to use steering,throttle and brake to execute our planned trajectory.