Final semester project for the course EL7008: Procesamiento Avanzado de Imágenes.

The idea is to create simple 2D occupancy grid maps & semantic maps using a monocular camera and semantic segmentation on U-Net.

1. Clone this repository:

git clone https://github.com/gonzal0lguin/monocular-2d-mapping.git

2. Clone the official Panther repository and replace the panther_macro.urdf.xacro file from the robot's description.

3. Install ROS dependencies:

cd ros/dev_ws
sudo rosdep install --from-paths src --ignore-src -y -r

4. Compile:

catkin_make

5. Download the pre-trained model and add it to:

segmentation_models/models/UNet/checkpoints

To start the simulation and vision node, run:

source setup.sh
roslaunch gazebo_sim city.launch worldname:=<name-of-world>

For Occupancy grid mapping, run:

rosrun mono_perception mapper

You can save the map with:

rosrun map_server map_saver map:=/global_occ_grid -f <your-map-name>

For semantic mapping, run:

rosrun mono_perception semantic_mapper

You can save the map as a gray-scale image with:

rosrun mono_perception save_sem_map.py