Skip to content

self driving car using Reinforcement Learning (Q- Learning)

Notifications You must be signed in to change notification settings

pparas007/Self-Driving-car

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 

Repository files navigation

Self-Driving-car

self driving car using Reinforcement Learning (Q- Learning)

The AI car senses the environment(i.e. its distance from the goal, the distance and presence roadside boundaries etc.), takes actions (go sraight, left, right) depending on the situation and receives rewards from the environment (+5 if car reaches the goal, -1 if car hit the boundaries, -2 if car hits the roadside, +0.5 if car's distance from goal is less than its distance at last instance or -0.1 otherwise i.e. negative reward for time spent) Using Reinforcement learning(RL) car learns from its mistakes appropriate and less appropriate actions to take while in particular situation. Q-Learning is the type of RL algorithm used in this project. Experience Replay is the additional technique used to allow AI to learn from the 'rare experiences'. This technique stores the past experience in the memory and learn from it by sampling random entries from it.

Currently the car can tackle a path as complex as shown in the screenshot uploaded. Further development can be done using more advance techniques such as reward engineering, addnig additional sensory inputs such as 'time spent' starting from the inintial point which will compail car to search for the shortest path.

Environment: Python, Pytorch, Kivy(for UI)

References: Udemy AI course

Email: paras.v.prabhu@gmail.com

About

self driving car using Reinforcement Learning (Q- Learning)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages