Moborobo_GMapping

Moborobo_SLAM

To access the last year’s work:

https://github.com/kaantuncer/Moborobo-Project/

Important! If you face cmake errors while building, install the package that written in the log

Also Orin cannot booted with SD card since it does not have jetpack image

Which changes are made?

FOR RSLIDAR:

• Run these commands in your workspace:

• git clone https://github.com/RoboSense-LiDAR/rslidar_sdk.git
  cd rslidar_sdk
  git submodule init
  git submodule update
  
  sudo apt-get update
  sudo apt-get install -y libyaml-cpp-dev
  sudo apt-get install -y  libpcap-dev  
  

• For more info, you can visit: https://github.com/RoboSense-LiDAR/rslidar_sdk

FOR MOBOROBO GMAPPING:

• rewriten line 17 in navigation.launch (node pkg gmapping part)
• removed robosense_16.dae geometry part in RS-16.urdf.xacro file (line 20-22)
• for rslidar we set the parameters for ethernet connection as:
  • connect the ethernet (yellow thick cable)
  • enter the wired settings from top-right menu
  • set the ipv4 method as manual
  • enter address as 192.168.1.102
  • enter netmask as 255.255.255.0
• some parameters are changed and added to the config.yaml file such as:
  • change-> send_packet_ros: true (line 6)
  • change-> lidar_type: RS16 (line 10)
  • add-> frame_id: rslidar (line 13)

FOR MOBOROBO SLAM WITH “slam-toolbox”:

On robot:

• sudo apt install ros-noetic-slam-toolbox
• Download the zip file in the link: https://github.com/SteveMacenski/slam_toolbox/tree/noetic-devel.

• You need to extract the zip file inside the on_robot/src directory

• sudo apt install ros-noetic-pointcloud-to-laserscan
• Run this command in your workspace/src (on_robot/src): git clone https://github.com/ros-perception/pointcloud_to_laserscan.git --branch indigo-devel --single-branch
• Go to the file Moborobo-Project/on_robot/src/pointcloud_to_laserscan/src/pointcloud_to_laserscan_nodelet.cpp and change the line 241 as : PLUGINLIB_EXPORT_CLASS(pointcloud_to_laserscan::PointCloudToLaserScanNodelet, nodelet::Nodelet); (inspired from https://github.com/ros-perception/depthimage_to_laserscan/issues/31)

Simulation:

• commented out line 17-23 in navigation.launch (node pkg gmapping part)

  • For execution:

    • sudo apt install ros-noetic-slam-toolbox
    • cp /opt/ros/noetic/share/slam_toolbox/config/mapper_params_online_async.yaml catkin_ws/src/moborobo/config/
    • roslaunch slam_toolbox online_async.launch params_file:=./src/moborobo/config/mapper_params_online_async.yaml use_sim_time:=true 2> >(grep -v TF_REPEATED_DATA buffer_core)

FOR ZED2 CAMERA

• sudo apt install ros-noetic-rviz-imu-plugin
  sudo apt install ros-noetic-rtabmap
  sudo apt install ros-noetic-rtabmap-ros
  sudo apt install ros-noetic-depthimage-to-laserscan
  
  cd /your_workspace(on_robot)/src
  git clone --recursive https://github.com/stereolabs/zed-ros-wrapper.git
  cd ..
  rosdep install --from-paths src --ignore-src -r -y
  catkin_make -DCMAKE_BUILD_TYPE=Release
  source devel/setup.bash
  
  cd /your_workspace(on_robot)/src
  git clone https://github.com/stereolabs/zed-ros-interfaces.git
  cd ../
  rosdep install --from-paths src --ignore-src -r -y
  

• Before catkin_make you should remove the zed-ros-interfaces file which is inside the zed-ros-wrapper file

• catkin_make -DCMAKE_BUILD_TYPE=Release
  source devel/setup.bash
    
  cd /your_workspace(on_robot)/src
  git clone https://github.com/stereolabs/zed-ros-examples.git
  cd ..
  rosdep install --from-paths src --ignore-src -r -y
  catkin_make -DCMAKE_BUILD_TYPE=Release
  source devel/setup.bash
  

• Run this command to test to see whether it works or not: roslaunch zed_display_rviz display_zed2.launch

FOR ROAD DETECTION

PyTorch Installation:

• Download the file torch-2.0.0+nv23.05-cp38-cp38-linux_aarch64.whl from the website https://developer.download.nvidia.com/compute/redist/jp/v511/pytorch/
• Run these commands directly

  export TORCH_INSTALL=~/Downloads/torch-2.0.0+nv23.05-cp38-cp38-linux_aarch64.whl
    
  pip install --upgrade pip
  pip install numpy==’1.16.5’
  pip install --no-cache https://developer.download.nvidia.com/compute/redist/jp/v511/pytorch/torch-2.0.0+nv23.05-cp38-cp38-linux_aarch64.whl
  

MobileSAM

• cd your_workspace(on_robot)/src/moborobot/scripts
  git clone https://github.com/ChaoningZhang/MobileSAM.git
  

• There is a file named mobile_sam inside the cloned repository
• Cut the mobile_sam folder and paste it to the scripts file (one level higher)
• Run the python script python3 path/to/python/script/road_detector.py

FOR RUNNING GPS:

• pip install pyserial
• sudo apt install ros-noetic-nmea-msgs

• Need to be sure that gps is connected with usb port

• if permission denied while running the py script:

  • sudo su
  • //type your password
  • cd /
  • cd dev
  • chown username ttyUSB0
• rosrun moborobo locoSysPublisher.py for publishing the gps data
• rosrun moborobo locoSysSubscriber.py for obtaining and reading the data (or you can simply run rostopic echo /moborobo/mygps)

Simulation: sudo apt-get install ros-noetic-hector-gazebo-plugins

PATH PLANNING:

How to Run?

• run this command in catkin_ws before launch: source devel/setup.bash
• for launch: roslaunch moborobo gmapping_room.launch 2> >(grep -v TF_REPEATED_DATA buffer_core)