 Maxime Chevalier-Boisvert
|
eb04bf87b5
Testing one-hot encoding
|
6 gadi atpakaļ |
 A. Baisero
|
f4f05f8b7b
fix sneaky view-limit bug in Grid.process_vis() (#34)
|
6 gadi atpakaļ |
|
A. Baisero
|
02bc4b8cf8
fix Grid.rotate_left for non-square grids (#33)
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
52eef49451
Added mission text box to human render window
|
6 gadi atpakaļ |
|
A. Baisero
|
0be0e71ba3
Implement Lava tiles and LavaCrossing environment (#7) (#29)
|
6 gadi atpakaļ |
 saleml
|
5b2053822b
field of view coordinates (#25)
|
6 gadi atpakaļ |
 d3sm0
|
f1a2080a32
Classical env and wrappers (#6, #13, #22) (#24)
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
8518802414
Made renderer import lazy to avoid PyQT dependency.
|
6 gadi atpakaļ |
|
saleml
|
341c80acd7
allow checking if a grid contains an object defined by a pair (None, object name) (#18)
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
7acd1ea326
Added timeout to place_obj and place_agent
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
c53358bf8f
Fixed bug in environment string representation
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
b6ffc9ba38
Added colored floor tile the agent can walk over
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
e270e76ee5
Added object position tracking
|
6 gadi atpakaļ |
 Lucas Willems
|
ea6416989f
Add `_rand_float` method (#10)
|
6 gadi atpakaļ |
|
Lucas Willems
|
c125ca7998
Minimum reward when success is now 0.1 (#9)
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
c99822121e
Added reward penalty based on number of time steps taken
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
636226f90e
Renamed wait action to done
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
0ad70c61dd
Added _rand_subset method to MiniGridEnv
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
40f29632c2
Added ability to render observations at any resolution
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
342b3c96f1
Fixed bug in place_agent
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
5d88cb4376
Doors can now be closed back after they are opened
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
4f4992265b
Fixed issue wrt agent view size pointed out by Anirudh
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
2190ce0b59
Full map correctly highlights cells visible to the agent
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
6b513362e6
Added gen_obs_grid method which outputs visibility mask
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
a6ece826ce
Removed old visibility code
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
76b43b7534
Added DIR_TO_VEC array. Agent position is now a numpy array.
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
146fd10741
Made reward_range the same for all environments, rewards are in [0, 1]
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
fe469bb1cc
Added MiniGrid._rand_color method
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
30c97ffc57
Added _rand_bool() method to MiniGrid
|
6 gadi atpakaļ |
|
Maxime Chevalier-Boisvert
|
d562fb8f82
Fixup. Default random seed should be fixed, deterministic.
|
6 gadi atpakaļ |
