Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Overhead camera view with an actual agent #15

Open
mathfac opened this issue Apr 3, 2019 · 2 comments
Open

Overhead camera view with an actual agent #15

mathfac opened this issue Apr 3, 2019 · 2 comments
Labels
enhancement New feature or request

Comments

@mathfac
Copy link
Contributor

mathfac commented Apr 3, 2019

author @dhruvbatra: [I believe this feature should be implemented in habitat-api not habitat-sim but please correct me if I'm mistaken]

Feature request: we should add an overhead camera sensor so we can see the agent move around like a player in a game.

See for example:
https://raw.githubusercontent.com/StanfordVL/GibsonEnv/master/misc/husky_camera.png
https://raw.githubusercontent.com/Unity-Technologies/obstacle-tower-env/master/banner.png

This could be an additional "sensor" that is attached to the agent, but the observations from this sensor are not used by the agent (the agent only uses ego-centric camera).
@mathfac mathfac added the enhancement New feature or request label Apr 3, 2019
dhruvbatra pushed a commit that referenced this issue May 10, 2020
@msavva
Fixes and improvements to benchmark script (#15)
ca01a85 
* Better multi-proc benchmark

* add instruction to run benchmark

* add rationale for picking fastest run when benchmarking

* fix output alignment, add argparse parameters for resolution, nprocs, and enabling semantic sensor

* add default test scene

* add automatic downloading of test scene data
@mathfac
Copy link
Contributor Author

mathfac commented Oct 27, 2022

@aclegg3 do we have this functionality already in master? Thanks.

@Skylion007
Copy link
Contributor

Not quite. The main thing we need is to automatically compute the appropriate camera pose based on the agent's bbox.

cpaxton added a commit that referenced this issue Jan 17, 2023
@cpaxton
Change ROS code to enable data collection (#15)
a068fa9 
* update data and collect some real robot data - involves changing ROS
package paths so that it will work

update collect data script - add some debug information

add data examples for ros teleop

update setup and add new robot data collection tools

add two cameras config for debugging

renaming the data tools

fix urdf path

add fix to data

add some missing features to ros Camera setup

get camera pose and save it in the recorder

fix a bunch of issues with missing code here

setting up real robot data collection tools

make it easier to see camera info when you're having issues getting it
to stick

update data collection for the robot + make sure we can get different
poses

add launch files for data labeling

* update - black formatting

* reformat xbox controller examples

* making sure rotate-images script comes up properly

* upate configuration for stretch-ros labeling tool

* add stretch pregrasp + update the visualizations for rviz labeler

* add some debugging information and tune pre-grasp training configuration

* data tools reformatted

* remove dead or unused or unfinished code and do some cleanup

* update stretch xbox - explicit imports

* update point cloud to remove mutables

* change arrays to tuples

* update folder structure + move the teleop code

* black formatting for pt cloud code
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
enhancement New feature or request
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@mathfac @Skylion007