- In terminal 1, run the following command to start realsense with ros, with the option to publish the pointcloud:
roslaunch realsense2_camera rs_aligned_depth.launch filters:=pointcloud
Check that the package react_inria
is installed and go to the folder react_inria/launch
.
- To provide the initial pose of the part in simulation, in terminal 2, while in the
react_inria/launch
folder:
roslaunch ./react_learning.launch use_sim_time:=true
And follow the instructions to generate a new pose and save it.
- Run the localization node in simulation, in terminal 2:
roslaunch ./react_localizer.launch use_wMo_filename:=$HOME/.ros/transform_wMo.yaml use_oMo_filename:=$HOME/.ros/transform_oMo.yaml use_aligned_depth:=false use_sim_time:=true
- In terminal 3, launch the script publishing the frame used in the demo, in
agimus-demos/script
folder:
./publish_frames.py
- To provide the initial pose of the part on the real robot, in terminal 2, while in the
react_inria/launch
folder:
roslaunch ./react_learning.launch use_cad_model_filename:=/root/devel/src/react_inria/config/Plaque-UR3.react
And follow the instructions to generate a new pose and save it. This needs to be done every time the part is moved in regards to the robot base.
- Run the localization node, in terminal 2:
roslaunch ./react_localizer.launch use_wMo_filename:=$HOME/.ros/transform_wMo.yaml use_oMo_filename:=$HOME/.ros/transform_oMo.yaml use_aligned_depth:=true use_cad_model_filename:=/root/devel/src/react_inria/config/Plaque-UR3.react
To use the neural network version of the localizer :
roslaunch ./react_localizer.launch use_wMo_filename:=$HOME/.ros/transform_wMo.yaml use_oMo_filename:=$HOME/.ros/transform_oMo.yaml use_dnn_localizer:=true use_cad_model_filename:=/root/devel/src/react_inria/config/Plaque-UR3.react
- In terminal 3, launch the script publishing the frame used in the demo, in
agimus-demos/script
folder:
./publish_frames.py