How could a robot learn to catch a ball thrown in its direction? This was the main question we aimed to tackle for our senior design project. Our robot was a XY style plotter robot with a cup attached to a gantry plate. To create a controlled projectile, we constructed a wooden ramp to roll ping pong balls down. The robot's goal was to go to the landing position of the ball before it landed. It used Q-Learning with a state representation of (x,y,velocity) of the ping pong ball and received a reward based on the Euclidean distance between the ball and the gantry plate. The ball position was determined by a camera above the ramp and two laser pointers measured the velocity. A Unity simulation of our enviornment was used to train the agent then transfered to the real life robot.
-
Notifications
You must be signed in to change notification settings - Fork 2
bryonkucharski/robot-catcher
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
About
Senior Design project to teach a robot to catch a ball using reinforcement learning
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published