v3.7

myGym 3.7

changelog:

new robots (Nico, Tiago, HSR)
new workspaces (human collaborative, Tiago table, Nico table)
new algorithms for multi-step training
visualization of multiple tranings in one graph
sim2real for Nico robot
new compositional rewards

myGym 2.2.1

Changelog:

• fixed IK bug with end effector and gripper navigation
• fixed bug with observation space
• languade module WIP
• new humanoid robot Thiago
• new algoritms for multi-policy - MultiACKTR

myGym 2.1.2

CHANGELOG:

• new visualization of training progress for multiple runs with variability - visaversucces.py
• fixed bugs with multi step rewarding
• new multi network algorithm - multiPPO2
• new task with diverse subgoals - Pick and rotate

myGym 2.1

Changelog:

• new URDF with robot automatic robot gripper annotation
• new robot upload and gripper control
• magnetic and mechanic gripper operation
• new tasks for multi step manipulation
• keyboard control of robots in test.py
• new observation definition
• robot force and speed control in config

myGym 2.0

Changelog:

• Tested support for Tensorflow 2.0
• Success training of PNP tasks
• Support for multi-step tasks
• Ability to train multiple networks within one tasks

myGym 1.5

Changelog:

• Faster implemetation of the training algorithms
• Novel tasks - press,turn, switch
• Better evaluation of the results
• Novel reward functions
• Template for own implementation of NN

myGym 1.0

First release of myGym toolbox