ROS 2 and its installation went smoothly following the instructions: https://docs.ros.org/en/foxy/Installation/Ubuntu-Install-Debians.html
Gazebo IGN had problems when trying to install a VM running Ubuntu, the problem was finally the vmware gpu driver when the host had a Quadro RTX 3000, the problem disappeared when the gpu got a native driver without vm. On the other hand, ignition was not needed in this project in the end.
The installation of Gazebo 11 went smoothly following the instructions: https://emanual.robotis.com/docs/en/platform/turtlebot3/simulation/#gazebo-simulation and https://ubuntu.com/blog/simulate-the-turtlebot3
Building the robot was easy, and things related to it made it easy, for example, the ssh connection between devices in the same WLAN and other similar tricks. eg robot control with master keyboard and robot in wireless mode.
this sprint i installed the camera system using the following guide: https://github.com/ros-perception/vision_opencv/tree/ros2/cv_bridge with opencv and v4l2 camera. getting the camera was simple but colcon build takes a lot of time with sudo upgrades etc. so lots of waiting when working with these external sensors for the rasp pi.
autonomous driving was working with physical robot with slam toolkit and nav2 where u can point on the map a target for the robot to move to. getting it working was also quite simple and didnt require anything extra special.
https://github.com/Le36/ros2-mapper/blob/main/documentation/tutorials.md
we created guides for ourselves to follow in the above link. it proved quite useful when working as a team.
so far no programming at all so there isnt much to report etc, just basic installing utilities.
this sprint i created the memory node system for the robot. it was very straightforward process following ros2 tutorials.
i used https://docs.ros.org/en/foxy/Tutorials/Writing-A-Simple-Py-Publisher-And-Subscriber.html this guide for it
Writing a simple publisher and subscriber (Python)
basically the system creates a topic it listens to and whenever data is published to that topic, it will be added to sqlite db
system could be extended this way easily in a few different ways. you could just add unique topic for all the db operations or have one topic handle them with different commands in that range.
what needs to be figured in the next spring is how to publish data from the db, maybe there is a need to refactor this system into service and client rather than publisher and subscriber system
surprisingly there were no really troubles when developing this node. the tutorial is very straightforward and since using sqlite u can have db up and running in 10 seconds
looking forward to next week, this sprint was more interesting as there was something to do programming wise albeit how lite it was
ok this was the big sprint for us. basically in this sprint we are supposed to finish the product. at the moment when writing this i can say that we are very close!
first i started the sprint with new db schema after customer gave us the requirements for it. it changed a bit along the sprint but here is the final version of it:
,----------------------.
|history | ,----------------------.
|----------------------| |qr_codes |
|id: int | |----------------------|
|center_x: float | |center_x: float |
|center_y: float | |center_y: float |
|center_z: float | |center_z: float |
|normal_vector_x: float| |normal_vector_x: float|
|normal_vector_y: float| |normal_vector_y: float|
|normal_vector_z: float| |normal_vector_z: float|
|rotation_w: float | |rotation_w: float |
|rotation_x: float | |rotation_x: float |
|rotation_y: float | |rotation_y: float |
|rotation_z: float | |rotation_z: float |
|time: timestamp | |id: int |
|pk: int | `----------------------'
`----------------------'
,--------------.
|sqlite_master |
|--------------|
|type: text |
|name: text |
|tbl_name: text|
|rootpage: int |
|sql: text |
`--------------'
so there is the required history system and the current qr codes that are supposed to be found in the stage.
making this part of the project was just extending the previous memory node that i created.
team refactored the node later on a little and integrated it to exploration node. this was also complicated because team had problems with the exploration node, also qr codes was changed to aruco markers.
i tried to help with commanding the robot, but the system was confusing for me.
https://automaticaddison.com/navigation-and-slam-using-the-ros-2-navigation-stack/
i tried to follow this guide and somehow modify it, but ended up not getting anything working really.
there were all kinds of problems, even my robot died at one point.
i also created some gazebo simulation worlds for testing purposes, which was fairly straight forward without major issues. at some points the team got the automatic exploration working in these simulations, but not on the physical robots.
overall this sprint was the biggest challenge, and it showcased the typical issues you could face in a project like this.
this is because almost everything we have made is based on some other open source project etc. so its very likely that even we dont understand everything we have made and refactored into our project. there isnt really anything you could say that we fully made ourselves and that is just for the nature of the project.