Commits · master · KiloBots / kbMazerunner

May 13, 2016
- removed paper. · a92cb388
  Gregor Gebhardt authored 8 years ago
  
  a92cb388
Mar 15, 2016
- final paper changes · 00c55b69
  Daniel Kauth authored 9 years ago
  
  View commits for tag IP15 IP15
  
  00c55b69
Mar 10, 2016
- some more small paper changes · a1ea3c75
  Daniel Kauth authored 9 years ago
  
  a1ea3c75
- minor changes to the paper · 01a5e403
  Daniel Kauth authored 9 years ago
  
  01a5e403
- added final reward data to the paper · f2c0d3bf
  Daniel Kauth authored 9 years ago
  
  f2c0d3bf
Mar 09, 2016
- wrote results and conclusion sections · cc79f296
  Daniel Kauth authored 9 years ago
  
  cc79f296
- fixed bug where policy parameters are not set · f2a35bbf
  Daniel Kauth authored 9 years ago
  
  f2a35bbf
Mar 08, 2016
- rewards are now saved · 87b7c1ba
  Daniel Kauth authored 9 years ago
  
  87b7c1ba
- added object movement policy section · 8110c5b5
  Daniel Kauth authored 9 years ago
  
  8110c5b5
Mar 07, 2016
- Merge branch 'learning-daniel' into 'master' · 8d2160cc
  Gregor Gebhardt authored 9 years ago
  
  Learning daniel See merge request !1
  8d2160cc
Mar 02, 2016
- added simulator section · 013e63e5
  Daniel Kauth authored 9 years ago
  
  013e63e5
Feb 28, 2016
- added AC-REPS sections · f61c78b9
  Daniel Kauth authored 9 years ago
  
  f61c78b9
- made it possible to restart learning without restarting python (continueLearning=False) · 10eee414
  Daniel Kauth authored 9 years ago
  
  10eee414
Feb 27, 2016
- added first version of paper (mostly just copied paper from part 1 and removed parts) · da221657
  Daniel Kauth authored 9 years ago
  
  da221657
Feb 26, 2016

refactoring · 56d3ede4

Daniel Kauth authored 9 years ago

separated learning and moving through maze
maze policy is now provided by the learner (for now just following a list of points)

56d3ede4

Feb 03, 2016
- testPolicies gives reward / fixed plotting of policy · a79b4f3b
  Daniel Kauth authored 9 years ago
  
  a79b4f3b
Feb 01, 2016
- added learner and working policy for the scenario where the object is pushed in any direction · 70e45419
  Daniel Kauth authored 9 years ago
  
  70e45419
Jan 27, 2016
- changed learner for object pushing scenario · aeb54e7d
  Daniel Kauth authored 9 years ago
  
  aeb54e7d
- fixed kernel weighting and value function · 45693b01
  Daniel Kauth authored 9 years ago
  
  45693b01
Jan 25, 2016
- added simple scenario with multiple kilobots · c82161d1
  Daniel Kauth authored 9 years ago
  
  c82161d1
Jan 20, 2016
- some refactoring · fd52648e
  Daniel Kauth authored 9 years ago
  
  fd52648e
Jan 19, 2016
- kilobot positions kernel is now much faster · 5efea273
  Daniel Kauth authored 9 years ago
  
  5efea273
- small changes to object pushing scenario · 42770f50
  Daniel Kauth authored 9 years ago
  
  42770f50
Jan 17, 2016
- Merge branch 'learning-daniel' of... · 1281814e
  Daniel Kauth authored 9 years ago
  
  Merge branch 'learning-daniel' of git.ias.informatik.tu-darmstadt.de:kilobots/kb_mazerunner into learning-daniel
  1281814e
- changed scenario to learn to push an object without a maze · 0610944c
  Daniel Kauth authored 9 years ago
  
  0610944c
- fixed kernel over kernel implementation · cd9953c3
  Daniel Kauth authored 9 years ago
  
  cd9953c3
Jan 12, 2016
- added and used kernel over kernel · e949caeb
  Daniel Kauth authored 9 years ago
  
  e949caeb
Jan 08, 2016
- removed FeatureFunction (Kernel is now used instead) · 4193091c
  Daniel Kauth authored 9 years ago
  
  4193091c
Dec 12, 2015
- added automatic saving of plots, policies + samples (every 5 iterations) and parameters · f0ab1b0d
  Daniel Kauth authored 9 years ago
  
  f0ab1b0d
- Merge branch 'learning-daniel' of... · 362b1c9f
  Daniel Kauth authored 9 years ago
  
  Merge branch 'learning-daniel' of git.ias.informatik.tu-darmstadt.de:kilobots/kb_mazerunner into learning-daniel
  362b1c9f
- made getRandomSubset more robust · f3ba466a
  Daniel Kauth authored 9 years ago
  
  f3ba466a
Dec 09, 2015
- added normalization to getRepresentativeRows · aeb5fdc0
  Alexander Hendrich authored 9 years ago
  
  aeb5fdc0
Dec 04, 2015
- "fixed" LinAlgError when using pinv · 7f5ce514
  Daniel Kauth authored 9 years ago
  
  reduced memory consumption added epsilon-greedy exploration
  7f5ce514
Dec 02, 2015
- added methods to save and load the policy · 2e61d784
  Daniel Kauth authored 9 years ago
  
  labyrinth now mostly works (gamma = 0.95)
  2e61d784
Nov 27, 2015
- fixed lstsq error by using pinv instead · 6fc94371
  Daniel Kauth authored 9 years ago
  
  6fc94371
Nov 25, 2015
- added function to plot the value function · 8be9093c
  Daniel Kauth authored 9 years ago
  
  8be9093c
Nov 24, 2015
- simple light movement example now works robustly with binary and continuous rewards · a1b1f058
  Daniel Kauth authored 9 years ago
  
  a1b1f058
- renamed evaluate() to getMeanAction() · ee55b1f9
  Daniel Kauth authored 9 years ago
  
  fixed sampleActions() to work when called with only one state
  ee55b1f9
- made computation of representative samples much faster · 8f7d3113
  Daniel Kauth authored 9 years ago
  
  8f7d3113
- fixed division by 0 · 1da71d44
  Daniel Kauth authored 9 years ago
  
  1da71d44