Robot navigation in outdoor mapped environment
I am working on a bunker robot from agilex. With the help of the Ros navigation stack, I am controlling the robot autonomously in an outdoor mapped environment. I have encountered a couple of errors
1) The robot is unable to localize itself at the exact point on the map because I place the robot in different positions before launching the navigation file and it always remains at the same point regardless of its actual position on the map.
2) Sending goal coordinates to the robot moves the robot but not in a very efficient manner. it doesn't move very smoothly but it rotates, stops, thinks, moves a bit, and then repeats the same steps again. it never reaches the exact point.
3) The robot-predicted path to the goal changes every while which I don't understand because the environment doesn't change.
4) I received the following error though I have set the global frame to /map and also checked it for /odom, the nature of the error doesn't change.
You need to explain a bit more. Are you using SLAM or a local planner? Which one? Does a lidar function well in your outdoor environment (this is unusual)? Are you trying to create a map, or use an existing map? It matters because the optimizations tend to be different.
Please edit your description using the "edit" button near the end of the descrption.
I am building the map using the hector_slam package to build my map. Kindly let me know if the gmapping package results in a more effective mapping of the environment. Lidar functions well in an outdoor environment as I have the boundary walls in the environment but the area is big
I have not used hector_slam, but your point (1) confuses me: most of these slam algorithms assume your robot is at map (0,0) at the start. As it builds a map, it will calculate relative map poses from that reference point. How can there be an "actual position on the map" when there is no map (yet)?
Also, with basic slam, I don't think that most people try to give navigation goals before there is an established map to use - instead they use a joystick (or some other human-based control) to move the robot around while allowing it to collect lidar data, which is then used to build a map. There may be other packages that do "autonomous exploration", but I don't see that feature in the hector_slam documentation.
1) Existing map: I am using an existing mapped environment that I have mapped with the hector slam package. Now I just want to navigate in the mapped territory..The mapping was done with help of an RF transmitter.
2) By localization I mean, when I launch the navigation file, the robot is always at the same point regardless of the actual position of the robot in the real mapped environment.
3) I don't know what you mean exactly by "autonomous exploration" but I am using hector slam for mapping the environment. if you have some effective approach for this task in your mind, it will be highly appreciated if you share it. Thank you