Commit Graph

159 Commits

Author SHA1 Message Date
d8e7c4c80f Follow gym 0.25 spec 2022-12-13 20:13:53 +01:00
048ba027f3 Fixed clipping of entities on edge 2022-12-13 20:13:39 +01:00
35864d4b38 Fix: StateObservable sometimes having wrong low-bound 2022-12-13 19:50:57 +01:00
f421c92f83 Observation spaces now define dtype to be float64 2022-12-13 19:45:13 +01:00
164c72504c Pong 2022-12-09 17:05:27 +01:00
3bb3ffa3a0 More configurability 2022-12-09 17:05:06 +01:00
5afa8b22b2 Better path visualizations; more agent & path configs 2022-12-09 11:20:15 +01:00
78ac536bb9 Fixed RayObservable Collisions PreFilter for Rect 2022-12-08 20:27:36 +01:00
2032d2e91d Register to Gym as Columbus-v1 2022-12-08 19:53:32 +01:00
292b12c5a1 Config-configurable width / height and minimal docs 2022-12-06 19:11:12 +01:00
ede0f80cea New Unit: au (=36px) 2022-12-06 12:52:32 +01:00
1bf587c4da Allow units for 'start_pos' (initial agent position) 2022-12-06 12:25:20 +01:00
6d465c69c9 Minor cleanup and reorganized some class definitions 2022-12-06 12:16:42 +01:00
8132bb9321 Force Units in px to be integers (rounding down) 2022-12-06 12:09:19 +01:00
bde7869f97 Implemented Units, Updated README, Example file for Units 2022-12-06 12:00:45 +01:00
7302d59727 Extended README, added Columbus.pdf 2022-11-25 11:28:19 +01:00
f02163d88c MIT License 2022-11-13 20:07:00 +01:00
06bfcfe065 Adding the envs from my thesis 2022-11-13 20:00:15 +01:00
d8856e7dc9 Added predefined color mappers (based on tanh and atan) 2022-11-05 17:39:13 +01:00
4fede26967 Confidence-Ellipse: Back to showing 1sigma 2022-11-01 16:25:47 +01:00
1ef66a7674 Confidence-Ellipse now shows 95% interval (and fixed edge cases) 2022-11-01 16:14:56 +01:00
555a4780e3 Invalidate value-map upon reward-change 2022-11-01 16:14:28 +01:00
0fe5c35dda Small bug fix regarding aux_penalty 2022-10-26 17:40:54 +02:00
e5ac7e0b05 README: Documentation for ConfigDefined Example 2022-10-25 15:02:33 +02:00
fcd502d05b Made ColumbusConfigDefinedExample compatible with humanPlayer.py 2022-10-25 15:00:21 +02:00
df8b2d836e README: Typo 2022-10-25 14:53:53 +02:00
910acc2a15 More precise aux_penalties 2022-10-25 14:43:29 +02:00
71809b7374 Allow manually invalidating value_map 2022-10-24 10:08:14 +02:00
d73c7798b3 Fixed bug in value_func visualization 2022-10-19 10:41:49 +02:00
89e5f1369e Made Value-Func Rendering way more efficient 2022-10-17 23:00:01 +02:00
ba796a1d6e Fixed bug with value_func-renderer 2022-10-16 19:15:23 +02:00
e371d871e6 Fixed value_func- and path-rendering not working 2022-10-16 18:58:32 +02:00
70536b79e7 Allow manual reseting of path via arg to reset() 2022-10-16 17:51:05 +02:00
66c9509c27 Implemented Visualization of Value-Function 2022-10-15 20:54:16 +02:00
e09a588e77 Allow not clearing path on reset 2022-10-15 11:16:13 +02:00
61160c8595 term_next hack actually not necessary 2022-10-14 17:08:12 +02:00
83eb759b9c Fix: Fatal Bug, max_steps always triggering 2022-10-14 15:23:40 +02:00
811b2bb1e3 Configurable max_steps for env 2022-10-14 11:32:37 +02:00
da655346cd Fixed last Fix broken 2022-10-11 19:01:34 +02:00
e8c3b5dac0 Fixed Off-By-One for final reward 2022-10-11 18:51:45 +02:00
d4a1a35190 Fixed Bug with path_overlay 2022-09-23 23:11:03 +02:00
678d6f620a Fixed bug with path drawing 2022-09-23 22:22:51 +02:00
29f96b1e8c README: Fixed typos 2022-09-23 19:02:38 +02:00
6636d8e1e9 Made draw_path more robust 2022-09-22 21:40:56 +02:00
100e82bd0f Updated README 2022-09-21 12:11:58 +02:00
a02878c5ab Draw paths and 'terminate_on_reward' 2022-09-20 21:57:41 +02:00
fd69dd4a5f Draw Paths and LoopReward 2022-09-20 21:57:16 +02:00
0412afb69c Implemented loader for CustomConfigEnv for humanPlayer 2022-09-20 21:56:50 +02:00
74ebb6e250 Removed debug-print and typo 2022-09-17 09:15:48 +02:00
6e1d6a9fd3 README: humanPlayer.py does not support ColumbusConfigDefined 2022-09-16 11:55:36 +02:00