JackM323
cd870a4afc
Win Initialize fix
2026-08-21 13:44:45 +02:00
JackM323
f5c07edc0a
unfinished Readme changes
2026-08-13 09:11:16 +02:00
JackM323
12b8f80002
import order changed
2026-08-07 14:51:37 +02:00
JackM323
3e6e40f0c5
pretrain logic and small fixes
2026-08-07 14:32:35 +02:00
JackM323
a1eb7b8573
no robot coloring
2026-08-07 14:31:58 +02:00
JackM323
7200531af3
deleted leftover file
2026-08-07 14:31:25 +02:00
JackM323
f9a3e8ddba
learning tweaks
2026-08-07 10:42:28 +02:00
JackM323
405e3ad5f2
Callback restored
2026-08-06 16:15:29 +02:00
JackM323
346ec9e949
simulation physics fixed
...
fuck them physics
2026-08-06 15:18:06 +02:00
JackM323
523a4aea89
added Robot kinematic use for training
...
HUGE BUG -> SimManager physics broken (at least with Robot kinematics)
2026-08-06 13:48:08 +02:00
JackM323
c93c524a10
code reduction for training and eval
...
reward and everything else changed
again
wont be the last time
2026-08-05 22:44:05 +02:00
JackM323
15e0206739
adjusted rewards and requirements
2026-08-05 11:08:33 +02:00
JackM323
f684bf44b8
Evaluation Phase Usage
2026-08-04 22:29:30 +02:00
JackM323
101004ead1
reward ajustments
...
robot starts balancing without changing phase and farms alive bonus
fix ->external force and bigger height penalty
2026-08-04 22:17:33 +02:00
JackM323
402c20dfb5
env cleanup
...
rewards adjustment
pybullet logic contained in SimManager
2026-08-04 21:03:53 +02:00
JackM323
766f2855da
Unused Exploration Bonus
2026-08-03 22:58:27 +02:00
JackM323
9d2e001ea4
Higher Height penalty
...
falling behavior was slightly positiv
therefore the penatly was increased again
2026-08-03 22:52:13 +02:00
JackM323
4fdfadfd3f
stage requirements more dependened on distance
2026-08-03 22:50:24 +02:00
JackM323
59b49d3b99
harder stage requirements
...
the robot should master the stage before proceeding to the next
2026-08-03 22:39:01 +02:00
JackM323
d50d5ac14f
Metric stacking bug fixed
2026-08-03 22:30:30 +02:00
JackM323
acb3d671be
Reworked training
...
new reward/penalty system
learning phases with curriculum learning
new training parameters
cleanup of old code
better logging while training
multiple environments instead of robots (they could bumb into each other)
2026-08-03 22:27:26 +02:00
JackM323
5317ef1299
fixed env eval setup
2026-07-31 18:35:09 +02:00
JackM323
a3e46c3abf
updated evaluation code for previous env changes
...
outdated code from previous changes on env
2026-07-31 16:33:50 +02:00
JackM323
846fbfaaab
Readme reward/termination explanaition added
2026-07-31 15:56:11 +02:00
JackM323
42974fd994
Readme Updated
2026-07-31 14:50:59 +02:00
JackM323
01f993d727
new default values for training
2026-07-31 14:17:22 +02:00
JackM323
61cb0150f0
Training uses new Robot.py
...
new Metrics and SimManager for live training viewing
Robot.py usage added to machinelearning
2026-07-31 13:55:24 +02:00
JackM323
9c31de3c38
Added Gui (unfinished)
...
config into dataclass and enums
new Gui that includes settings
deleted GlobalVariables
small fixes (import, names...)
2026-07-30 22:49:53 +02:00
JackM323
b537677277
Complete Restructered Robot Code
...
Robot into its own Class instead of lose Global Variables that cause circular imports
StateClass usage instead of the old RobotState.py
New Input Class for Controller and randome intputs
2026-07-30 21:14:50 +02:00
JackM323
5448335b11
Text Metrics for Testing in the Training
2026-07-30 17:35:12 +02:00
JackM323
5d598b4d94
Urdf update with collision
2026-07-30 17:23:50 +02:00
JackM323
c5ca79a354
Machine Learning Trainer
...
Training environment to make a walk model for the hexapod
generated code that will be checked
2026-07-30 17:23:36 +02:00
JackM323
bec157a458
UTF-8 Symbol Causes Errors on Linux
2026-07-30 08:40:29 +02:00
JackM323
2b2125bfde
merge conflict and readme update
...
old code was uploaded in the initial upload
changed everything with working programm and updated the readme
2026-07-29 22:35:41 +02:00
JackM323
1e660fcdb6
Add copy methods, fix state imports, pass lock
...
Add copy() to PosArray, DegArray and RadArray to allow easy shallow copies. Change RobotState imports in idle.py and walking.py to use explicit 'from RobotState.RobotState import RobotState' to avoid import issues. In main.py create a threading.Lock and pass it into the controller thread args to prepare for synchronized access in controller_loop.
2026-07-29 22:17:28 +02:00
JackM323
fa6b0fd624
linux/windows venv build comms
2026-07-29 21:52:19 +02:00
JackM323
bc84f7e8eb
cleanup
2026-07-29 20:38:30 +02:00
JackM323
c34449aac5
Project Initialization
...
First Upload to Gitea
2026-07-29 19:23:45 +02:00
JackM323