- Our paper is available: Unified Understanding of Environment, Task, and Human for Human-Robot Interaction in Real-World Environments
- OS: Ubuntu20.04
- ROS noetic
- CUDA: 11.7
- Please refer this repository
mkdir -p <your_ros_ws>/src
cd <your_ros_ws>/src
git clone --recursive https://github.com/tamhome/tam_dynamic_map.git
catkin build
rosrun tam_dynamic_map recog_furniture_node.py
- Auto generation by recognize with OMNI3D
rosrun tam_dynamic_map create_world_model.py
- You can create/edit world model manually
- open map in your editor
world:
- id: white_table
type: table
pose: {x: 3.75, y: 0.72, z: 0}
scale: {x: 0.50, y: 0.75, z: 0.74}
# - id: object_id_as_you_like
# type: chair # load furniture templates in ./io/furniture_templates/<>
# pose: {x: 3.75, y: 1.72, z: 0} # you can edit object pose
# scale: {x: 0.50, y: 0.50, z: 1.00}
rosrun tam_dynamic_map load_world_model.py
- If you want to use
sample.yaml
as a world model, you have to setsample
as follows.
<arg name="world_model_name" default="sample"/>
<!-- world model 読み込み -->
<include file="$(find tam_dynamic_map)/launch/load_model.launch">
<arg name="world_model_name" value="$(arg world_model_name)"/>
</include>