Skip to content

We use visual data alone to learn a control policy for a robotic arm by observing expert video demonstrations. We implement and test several models, accomplishing an 85% success rate for a pick-and-place task.

Notifications You must be signed in to change notification settings

jeremy-collins/vroom

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

vroom

Learning Robotic Tasks from Video Observation

convlstm usage

checkout alantest branch git checkout alantest

Ensure checkpoints directory exists mkdir -p checkpoints

Data should follow this example structure for train and test directories: ./RoboTurk_videos/bins-Bread/test/demo_XXX_jointdata/frame_XXXX.npy ./RoboTurk_videos/bins-Bread/test/demo_demo_XXX/frame_XXXX.jpg

Modify hyperparameters in trainer_lstm.py

Run trainer script python trainer_lstm.py --folder `realpath RoboTurk_videos/bins-Bread` --name lstm --save_best True --dataset roboturk

About

We use visual data alone to learn a control policy for a robotic arm by observing expert video demonstrations. We implement and test several models, accomplishing an 85% success rate for a pick-and-place task.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published