1d encoder implementations #266
Conversation
…utonomous_f1tenth into 1d-encoder-implementations
…omous_f1tenth into 1d-encoder-implementations
…ARES/autonomous_f1tenth into 1d-encoder-implementations
There was a problem hiding this comment.
These should be added to gitignore
| odom[5] = odom[5] - self.offset[5] | ||
|
|
||
| limited_odom = odom[-2:] | ||
| # if self.firstOdom: |
There was a problem hiding this comment.
I know it looks ugly but this needs to stay for the real car odom to work. Can probably be changed to look nicer but logic is necessary.
|
|
||
| scan = create_lidar_msg(lidar, len(lidar_range), lidar_range) | ||
| self.processed_publisher.publish(scan) |
There was a problem hiding this comment.
Not all cases seem to produce a variable called scan so this could break the code.
| ############################################################## | ||
| ## TEMPORARILY OVERRIDING NETWORK CONFIG FOR TD3AE AND SACAE | ||
| ############################################################## | ||
| _,_,network_config = parse_args_from_file() |
There was a problem hiding this comment.
If network_config is assigned here it shouldn't be receiving the variables from parse_args too
| ##################################################################################################################### | ||
| # CHANGE SETTINGS HERE, might be specific to environment, therefore not moved to config file (for now at least). | ||
|
|
||
| # Reward configuration | ||
| self.BASE_REWARD_FUNCTION:Literal["goal_hitting", "progressive"] = 'progressive' | ||
| self.EXTRA_REWARD_TERMS:List[Literal['penalize_turn']] = [] | ||
| self.REWARD_MODIFIERS:List[Tuple[Literal['turn','wall_proximity'],float]] = [('turn', 0.3), ('wall_proximity', 0.7)] # [ (penalize_turn", 0.3), (penalize_wall_proximity, 0.7) ] | ||
| self.REWARD_MODIFIERS:List[Tuple[Literal['turn','wall_proximity','lin_acc'],float]] = [('turn', 0.2), ('wall_proximity', 0.6), ('lin_acc',0.2)] # [ (penalize_turn", 0.3), (penalize_wall_proximity, 0.7) ] |
There was a problem hiding this comment.
Not sure if it's just github showing it weird but if not this needs to be indented
There was a problem hiding this comment.
Is there a way to modularize the AE stuff? Seems a bit clunky putting all of it within init so every algorithm gets all these unneeded variables
There was a problem hiding this comment.
Update to latest submodule
There was a problem hiding this comment.
Need to ensure compatibility with latest version of cares_rl
No description provided.