4.9 movement, obstacle avoidance and navigation

• Now there were two development model (framework)
• The first, under the ROS for the robot body, a camera, a robotic arm and other development interface, the formation of many separate package
• Such apparatus is to classify the packet, sub-module development
  
• Second, the use of ROS MoveIt! Development kit, in this broad framework, developed for different robots and robotic functions
• This is according to different functions, interface development!
• The second seemingly coming into fashion! !
• Interface development! !
  
• Goal: to understand the principle, to see how to achieve movement and obstacle avoidance behavior without loading the map, the robot behavior design
• Does that mean you can not load maps loaded an empty map? (If not empty, only to a global wall, like a rectangle-like), and then specify the initial coordinates manually, you can perform local path planning, because the global map is empty, not equivalent to a global plan, in fact, global planning semantic map in LV
• Implementation: Given a virtual door position (position, distance and angle), the robot can walk, obstacle avoidance, adjust the position and attitude
• Using the ROS MoveIt! and Navigation Stack
• installation
• sudo apt-get install python-wstool

• STEP 2: Download the source code

source /opt/ros/indigo/setup.bash mkdir moveit cd moveit mkdir src cd src/ wstool init . wstool go https://raw.github.com/ros-planning/moveit_docs/indigo-devel/moveit.rosinstall wstool update cd ..

• STEP 3: Make sure MoveIt! dependencies are installed

rosdep install --from-paths src --ignore-src --rosdistro indigo -y

• STEP 4: Build MoveIt!

• Assuming you are in the moveit folder created above,
• catkin_make
• Such equipment is too slow
• Recommend this installation? ?
•  sudo apt-get install ros-indigo-moveit-full
• Since this does not look at the source code, so uninstalled
• According to the source, or to build it
• Download and install dependencies
• Copy the folder to the workspace, right next ROS
• plan
• The manual for your starting point
• Semantic map as a global planner
• To implement a manual robot teleoperation P3AT
• Then complete semantic navigation robot
• And then realized the door chase
• Achieve P3AT manually on their computers and find vontrol base, the only way to use the navigation stack, using an analog mobilesim
• View move_base code! Thinking improvement program, and ultimately on their computers to achieve, at a given point sequence, the robot can achieve local obstacle avoidance, and complete the path.
• amcl node    
• Look not start this node will not go wrong. If an error occurs, the transformation, so that the function of this node is positioned manually. zslocal
• Have reviewed globleplanner, movebase, navcore completed
• Continue to look at the code, analyze the need to modify where, and how to modify.
• Why can not receive other messages ROSARIA it?
• tf reasons and related research
• Research navigation stack and code changes
• Uninstall deb package, but want to return to install bfl third party dependencies, compile later found problems, or on all the dependencies installed it
• sudo apt-get install ros-indigo-bfl libasound2-dev libavahi-client-dev libavahi-common-dev libcaca-dev libdbus-1-dev libpulse-dev libsdl-image1.2-dev libsdl1.2-dev libslang2-dev libwebp-dev libwebpdemux1 libasound2-doc
• All relevant globalplanner and global costmap comments relevant code
• Because of this global project planner (semantic navigation module) is not complete, so the recovery module also commented that this module is not loaded
• FML, TM is too difficult, too much code Estimated build all pass, not to mention the debugging, but also quaternion! !
• Method Two: Use the map blank map, using false positioning, design their own global planner
• The above estimates and as difficult;
• Method three: writing their own zsbase
• But it also have to understand is okay on the basis movebase! !
• FML, too hard, to spend a day or two to see code!