Commit Graph

Select branches

Hide Pull Requests

debug_value-dice

dev-idm-videos

horner_scheme

idm-override

idm_upgrade

idm_upgrade_proj

main

overfit-a

save-videos

setup

smaller-convnets

tune-gail

icra23

iros22

Mono Color

• 06785236d4 fixing expert data combiner and adding feasibility checking with all options to util.collisions Arec 2021-10-21 06:59:21 -07:00
• 2d8928f2ae updating data processing scripts to output to the correct location Arec 2021-10-21 06:29:43 -07:00
• 05b31092f4 moving options policy to policies, commenting options image, and making the calls to train more flexible Arec 2021-10-21 05:59:28 -07:00
• 45a99978e4 adding discriminators to main folder, utilities to render a video from a saved model Arec 2021-10-21 05:33:05 -07:00
• da1fb11269 adding functions to process expert data across locations and tracks in intersimple environment Arec 2021-10-21 05:21:06 -07:00
• dcf8212028 adding comments to gail_options_image, combining environments for options gail, and fixing bug where last state is yielded in hl buffer Arec 2021-10-20 12:28:12 -07:00
• bef6d6db55 Merge branch 'main' of github.com:sisl/InteractionImitation Johannes Fischer 2021-10-20 12:06:28 +02:00
• 7b900d6260 Add exact two-stage collision checking method Johannes Fischer 2021-10-20 12:02:25 +02:00
• 73a160fa7a Fix /ray/ gitignore Johannes Fischer 2021-10-20 11:42:14 +02:00
• b61e4a5002 adding multi-circle collision checker Arec 2021-10-19 07:21:10 -07:00
• eae8c7f3f4 Fix mask computation ebuehrle 2021-10-19 09:35:14 +02:00
• 466e6b6ce7 No-mask options GAIL ebuehrle 2021-10-18 13:28:32 +02:00
• 62bfe797a4 adding function to load experts and combine them into single transition list Arec 2021-10-15 02:47:52 -07:00
• ffb16cfc31 updating function to process all expert data from track files, starting processing options policy from file Arec 2021-10-15 02:39:49 -07:00
• 415d607418 making the expert demonstration processor go through all agents in order when producing a single (default) trajectory file, using a randomized agent environment in optionsgail, starting function to process and store all expert data Arec 2021-10-13 11:20:40 -07:00
• 70e55327dc fixing flataction discriminator to convert to float beforehand, adding necessary forward calls in expert, adding Fire to video creator from model, and trying full run of options gail with new discrimination model Arec 2021-10-11 08:47:18 -07:00
• 01752fac12 adding model to append actions to encoded image state for discriminator, files to test different environment experiments, and a script to render an environment from a saved model Arec 2021-10-11 07:14:03 -07:00
• 244b85a06e Add larger expert dataset creation command Johannes Fischer 2021-10-08 18:23:59 +02:00
• 8a67d3536e Add ray directory to gitignore Johannes Fischer 2021-10-08 18:20:20 +02:00
• c59da3611c Use ray tune in gail Johannes Fischer 2021-10-08 18:19:08 +02:00
• 6d2ab54b6e Add callback to gail image random to report metrics Johannes Fischer 2021-10-08 18:18:42 +02:00
• 4b9a81080b Add missing * in **kwargs Johannes Fischer 2021-10-08 18:17:02 +02:00
• 59681cb16f Merge pull request #3 from sisl/refactor-sampling ebuehrle 2021-09-15 09:08:13 +02:00
• 4928458e08 Fix discriminator reward ebuehrle 2021-09-15 07:15:43 +02:00
• 40da84393c Update imitation version ebuehrle 2021-09-14 18:02:56 +02:00
• 183657dc36 Fix buffer bug, add test ebuehrle 2021-09-14 17:55:02 +02:00
• d2932374d9 Refactor sampling ebuehrle 2021-09-14 11:54:49 +02:00
• 454636665f Use ray tune tune-gail ebuehrle 2021-09-14 13:02:33 +02:00
• 763a7bb0d3 Merge branch 'main' of github.com:sisl/InteractionImitation Johannes Fischer 2021-09-14 11:01:24 +02:00
• deaef45943 Add test for discriminator ebuehrle 2021-09-14 07:17:43 +02:00
• 2280597db6 Add GAIL with random agent data Johannes Fischer 2021-09-13 19:42:41 +02:00
• 7eae74a7d8 Switch back to additive reward ebuehrle 2021-09-13 17:55:50 +02:00
• b0b358544f Merge pull request #2 from sisl/fischer/fusion_sample_methods johannes-fischer 2021-09-13 16:51:16 +02:00
• 9c7e6cef3a Rename action to option Johannes Fischer 2021-09-13 16:43:45 +02:00
• 9b8ceed9c9 Render only one episode Johannes Fischer 2021-09-13 16:31:24 +02:00
• 826c0fa219 Merge different methods to sample the policy and collect transitions Johannes Fischer 2021-09-13 15:50:49 +02:00
• f1ece358d7 Speed up collision check, assume 0 is fallback option ebuehrle 2021-09-11 20:59:32 +02:00
• 87ff3dbb93 Add cuda support, normalize actions ebuehrle 2021-09-11 14:47:35 +02:00
• e7b0aea427 run options gail ebuehrle 2021-09-10 16:15:34 +02:00
• 8ad7457159 Merge branch 'main' of github.com:sisl/InteractionImitation Johannes Fischer 2021-09-10 11:38:31 +02:00
• 544ea4d15a Allow variable horizon trajectories Johannes Fischer 2021-09-10 11:34:36 +02:00
• 9a107b165a Allow variable horizon trajectories Johannes Fischer 2021-09-10 11:34:36 +02:00
• b2b2abafa2 Fix typo Johannes Fischer 2021-09-10 11:34:13 +02:00
• 5ff4b42c0e Fix imports ebuehrle 2021-09-08 20:45:45 +02:00
• 3a6139286d Update .gitignore ebuehrle 2021-09-08 20:31:45 +02:00
• 50916aec05 Vanilla GAIL on rasterized observation ebuehrle 2021-09-08 20:28:18 +02:00
• 802d4a4301 Try lower image resolution ebuehrle 2021-09-08 20:26:54 +02:00
• f94ec9a4dc Add test for discriminator ebuehrle 2021-09-08 20:23:39 +02:00
• de5877aaad filling in available_actions, generate_plan, and feasible helpers Arec 2021-09-08 05:14:15 -07:00
• 26cde18be3 Add debug output and original implementation debug_value-dice Johannes Fischer 2021-09-08 13:52:21 +02:00
• 1fd0a71646 Draft Options GAIL ebuehrle 2021-09-07 19:22:49 +02:00
• d89e491b92 Remove debug print statement ebuehrle 2021-09-07 19:20:28 +02:00
• e9f09cacb7 Add option to render expert rollout ebuehrle 2021-09-07 19:19:53 +02:00
• 88e0b99d7e Add script for data generation ebuehrle 2021-09-07 19:18:59 +02:00
• a70907c0fd Copy over experiments ebuehrle 2021-09-01 16:05:46 +02:00
• 317d329765 Add shell script for value dice training Johannes Fischer 2021-08-06 18:53:32 +02:00
• 88b4466e57 MInor change in value dice loss, activate print statements, only do EITHER value OR policy update for each batch Johannes Fischer 2021-08-06 18:52:17 +02:00
• 025c71767f Change final value network activation to identity Johannes Fischer 2021-08-06 18:48:53 +02:00
• 66bfba3986 Minor formatting Johannes Fischer 2021-08-05 18:37:54 +02:00
• 6afb112277 Bugfix in value dice Johannes Fischer 2021-08-05 18:35:14 +02:00
• bf4c19a4d0 Add value dice ray config Johannes Fischer 2021-08-05 18:33:34 +02:00
• 8bce5d15f6 Restore train_epochs to 200 instead of 8 Johannes Fischer 2021-08-05 18:32:46 +02:00
• 40c55478f3 Bugfixes in valuedice Johannes Fischer 2021-08-04 20:53:03 +02:00
• 2224e2cd14 Merge branch 'main' of github.com:sisl/InteractionImitation Johannes Fischer 2021-08-04 19:44:04 +02:00
• 5c40de66fa Implement ValueDICE and some restructuring Johannes Fischer 2021-08-04 19:34:49 +02:00
• f9729b0a9d making expert data save s, a, sp. making dataloader also load batches thisway. renaming state to ego_state. converting path_x and path_y to single path variable. making number of samples for ray an argument. adjusting metrics, policy, and other functions to be able to handle this Arec 2021-08-04 09:45:36 -07:00
• 7ae01f73a2 AdVIL tests Etienne Buehrle 2021-08-04 16:41:18 +02:00
• cba42c6e4d Set default divergence to histogram based Johannes Fischer 2021-08-03 17:18:30 +02:00
• 5919a4e439 Use JS divergence in metrics Johannes Fischer 2021-08-03 17:04:47 +02:00
• 1ee46214a7 Implement jenson shannon divergence Johannes Fischer 2021-08-03 17:03:37 +02:00
• 1916a8fe69 Implement metrics and write to tensorboard summary at test time Johannes Fischer 2021-08-03 15:24:24 +02:00
• 98294e0c95 Merge branch 'main' of https://github.com/sisl/InteractionImitation into main Arec 2021-08-03 03:28:23 -07:00
• 943e8cda26 adding output directory to parse arguments Arec 2021-08-03 03:28:11 -07:00
• 6e524cf4b5 adding options for regularization and relative state masking via interaction graphs during data processing and experiment running. found 0.002 regularization on actions gives up to 3m of deviation with no collisions. added shell script to run ray experiments overnight Arec 2021-08-02 14:38:06 -07:00
• b869597717 extend comment on divergence Johannes Fischer 2021-08-02 19:15:02 +02:00
• f468b3b7a4 Implement histogram based kl divergence computation Johannes Fischer 2021-08-02 19:14:40 +02:00
• 9dd655bc75 test out kd divergence estimate based on CV-KDE (very slow) Johannes Fischer 2021-08-02 17:38:52 +02:00
• 6177b1f7e1 Add kl_cat Johannes Fischer 2021-08-02 11:12:24 +02:00
• 4ea4d42df7 Merge branch 'main' of github.com:sisl/InteractionImitation Johannes Fischer 2021-08-02 11:08:35 +02:00
• 281f7773c4 Implement kl divergence methods and tests Johannes Fischer 2021-08-02 11:04:03 +02:00
• 99f7df2e7c Add comment Johannes Fischer 2021-07-30 18:19:29 +02:00
• 367f72ec46 Imrove print output Johannes Fischer 2021-07-30 18:19:15 +02:00
• fb91ee1a62 Merge branch 'main' of https://github.com/sisl/InteractionImitation into main Arec 2021-07-30 05:17:43 -07:00
• 5fcd3f49da adding hyperopt tuning Arec 2021-07-30 05:17:38 -07:00
• bd5e854720 Add collision and avg velocity metrics Johannes Fischer 2021-07-30 13:35:42 +02:00
• e30d0ab1ba removing outliers from expert tracks Arec 2021-07-29 08:20:40 -07:00
• 3b4ef6ffb5 adding tool for visualizing acceleration distributions, and making nframes an arg Arec 2021-07-29 07:35:21 -07:00
• afc3719ab9 Fix bug with wrong argument order Johannes Fischer 2021-07-29 15:30:55 +02:00
• 3b623a1467 Add testing script for raytune experiments Johannes Fischer 2021-07-29 15:30:41 +02:00
• 35634fd2eb Separate experiment from main.py Johannes Fischer 2021-07-29 14:40:19 +02:00
• e47d69dbc1 changing relative state dim to 6 Arec 2021-07-29 02:54:41 -07:00
• fdbdc7f9f0 making data processing happen on front end, not on data loader. saving a ton of time Arec 2021-07-29 02:53:20 -07:00
• 09f77e0587 changing how tune reporting works so the scheduler doesnt break if itcant find cv loss. also fixed bug in config structure that was rendering impossible policies Arec 2021-07-28 09:55:11 -07:00
• 5a7090a21c Merge branch 'main' of https://github.com/sisl/InteractionImitation into main Arec 2021-07-27 14:32:52 -07:00
• 87aa19b86b getting hyperparameter tunning with ray tune working. updating default network with optimization and general parameters. Arec 2021-07-27 14:30:28 -07:00
• 2bd638c247 Test for batch data in test_deepsets_policy Johannes Fischer 2021-07-27 12:37:40 +02:00
• ca80fa19eb Merge branch 'main' of https://github.com/sisl/InteractionImitation into main Arec 2021-07-26 07:52:19 -07:00
• d9436daaba changing default network, making deepsets network choose output dimension appropriately, making bc config to be called by ray Arec 2021-07-26 07:52:11 -07:00
• 5e89a83c88 Fix tests Johannes Fischer 2021-07-26 16:35:45 +02:00
• 69359b5af3 periodically savingin out model and adding functionality to make identity Phi networks (for 0-dim NNs) Arec 2021-07-26 05:51:08 -07:00