-
Notifications
You must be signed in to change notification settings - Fork 57
/
Copy pathCHANGELOG
153 lines (131 loc) · 4.61 KB
/
CHANGELOG
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
## v0.3.4 (2022.7.25)
- Add Traj SAC and PPO policy
- Add MetaDrive evaluator
- Add MetaDrive TREX demo and doc
- Update Carla support to 0.9.10
- Fix bug of MetaDrive reset conflice with EnvManager
- Fix bug in DingEnvManager
## v0.3.3 (2022.6.5)
- Update readme, reorganize info
- Add MetaDriveTrajEnv and doc
- Add utils for MetaDrive
- Modify utils for macro env
## v0.3.2 (2022.4.24)
- Update banner logo
- Update to DI-engine 0.3, modify env properties
- Fix Basic policy bug
- Fix CILRS model bug
- Fix bugs in evaluator
- Fix bug in ppo config
- Fix bug in simple rl pipeline
- Add pre-train weights path in foc
## v0.3.1 (2022.2.25)
- Add MetaDrive Macro Env
- Add common models
- Add MetaDrive TopDown view PPO and SAC train
- Add Macro env SAC PPO train and eval
- Apply new distributed pipeline to simple rl
- Update simple rl doc
- Modify all PPO configs for new DI-engine
## v0.3.0 (2022.1.7)
- Add MetaDrive support (env, entry, doc)
- Add simulator selection when import
- Rename all common modules without 'carla'
- Add MetaDrive basic env training
- Add MetaDrive Macro env training and test
- Add all env registry
- Update DriveEnvWrapper
- Update logo
- Update doc, add metadrive guide and tutorial
- Reformat simple rl, all policies now have common entry(train eval test)
- Fix LBC planner bug
- Delete config helper
## v0.2.2 (2021.12.3)
- Add fail count and retry for benchmark collector
- Add LBC Image training phase 0 & 1
- Add comments for commonly used default config
- Add LBC Image train doc; scenario doc; save replay doc
- Update ci and setup
- Update simple rl env wrappers
- Update args in all Carla policy
- Modify all eps-greedy input into learn iterations
- Fix bug in CILRS model
- Fix typos in doc and comments
## v0.2.1 (2021.11.18)
- Add NoCrash Carla099
- Add bev dataset
- Add Carla AD Challenge scenarios and routes
- Add LBC data collection and BeV training doc
- Add Latent RL BeV and RL training doc
- Add validation in LBC training
- Update Benchmark datasaver
- Update VAE model
- Update LBC Image policy learn mode
- Fix bug in benchmark collector
- Fix visualizer not closed after done
- Fix not alived sensor killed bug
## v0.2.0 (2021.11.1)
- Update DI-engine to version 0.2.0
- Reformat CIL to CILRS
- Add CILRS dataset, model, policy and entry
- Add LBC data collection and dataset
- Add LBC BeV training entry
- Add simple RL demo doc
- Add logger and should eval for all evaluators
- Add enable save replay in DI-engine demos
- Update IL tutorial and CILRS demo doc
- Update Steer Noise wrapper
- Update Benchmark datasaver and collector
- Fix import bug in sensor helper
- Fix visualization bug
## v0.1.3 (2021.9.19)
- Update DI-engine to version 0.1.1
- Add serial evaluator for evaluation in training. It has tensorboard
- Update single and benchmark evaluator
- Update simple_rl entries. Adapt ding 0.1.1
- Modify models in simple_rl. Use modules in ding
- Modify env wrapper in simple_rl. Add inputs for discrete wrapper
- Update config system in RL. Move all without server into env and policy
- Update comple_config to adapt new system
- Add CICT eval
- Fix bug in BasicPlanner
- Fix bug in implicit train_sl
- Fix figure in doc
- Add auto doc in github pages
## v0.1.2 (2021.9.2)
- Rename 'orientation' to 'forward_vector'
- Redefine train, test, and eval in demo
- Move 'utils' in simple rl into core
- Add Latent RL demo, including env, DQN train, test and eval
- Add MPC controller with waypoints input
- Add AutoMPCPolicy, rename AutoPolicy to AutoPIDPolicy
- Add Vanilla VAE model
- Add API doc for MPC controller
- Add API doc for LPC Policy
- Update evaluator. It now returns success / success rate
- Fix a bug that walker controller may be identified as walker in CarlaDataProvider
- Fix birdview visualizer to show less channels
## v0.1.1 (2021.8.16)
- Update AutoPolicy: better performance in multi-lane maps
- Add bike avoidance and agent state in behavior planner
- Add CICT IL method: data, train, eval and test
- Add CICT doc
- Add LBC: eval and test
- Add LBC: doc
- Add common PID controller
- Add LBC planner
- Add transform obs for evaluator
- Add velocity, angular velocity and rotation in simulator's navigation and data provider
- Modify initialization of AutoPolicy
- Fix bugs in default config of all modules
- Add overview for Implicit Affordance doc
## v0.1.0.1 (2021.7.21)
- Fix import error in scenario files.
- Fix CIL running issues (collect/train/eval).
- Fix a bug in ResNetV2 model.
- Fix a typo bug in scenario manager.
- Remove useless data utils in CIL.
- Change some expressions in simulator API and casezoo doc.
- Add specific version of scipy in setup.py
## v0.1.0 (2021.7.8)
Initial project