In this lab you will use a graphical environment to test different motion planning techniques. You will use OMPL, the Open Motion Planning Library, which comes both for integration into existing projects as well as with a graphical interface to visualize planning. More information about OMPL can be found on the OMPL homepage via the link below.
Any problems discovered during the lab, and any clarifications we make to the lab instructions, will be announced here.
The following resources are available:
OMPL is started by executing
If you want to run OMPL from home you have two options. Either install it locally or run it remote via ssh. We cannot however offer support for these options and hence recommend that you carry out the lab in the PULs. When you start OMPL.app which is the graphical interface of OMPL you will see something like this:
Start OMPL.app. In this part we will take a look at some different types of motion planners and compare their approaches. OMPL comes with many resources in the form of maps and robots. For simplicity we will use 2D maps and robots although the gui always renders the image in 3D. To get 2D planning set the "robot_type" to "Rigid body planning (2D)". The mentioned resources are found in
Start with choosing a map that you think looks interesting and add a robot to it. Map files have names ending with _env.dae and robot files have names ending with _robot.dae. Before you can start planning you need to select a start and a goal location. Note that you can zoom in/out with the mouse wheel and if needed, rotate the map by dragging the mouse while holding a button pressed.
Now you can choose a planner out of the many existing to get your first plan in OMPL. After a plan is found you can choose to have it animated step by step or shown. You can also show states and edges between states to see which states and paths the planner generated.
For the rest of the lab, select two maps which you think will capture the different aspects of motion planning. Then use the following information to select at least 3 different planners that you run on the same problems. In the report, write which you have chosen and describe how they work in general terms. Also report your findings when it comes to plan quality, generated states and time used by the planner. Planner statistics can be found in the log window, started via Tools->Log window or CTRL+1. It is important that you add screenshots to the report so that the lab assistant can follow your reasoning. Note that random planners will generate different plans each time so don't forget to screenshot (alt-print sceen) before you move on. It is also possible and recommended to save the paths for the report.
If the planner takes considerably longer time than allowed by the time limit it may have crashed. Therefore, it is always a good idea to check the output in the logwindow.
When you hand in your results:
Page responsible: Jonas Kvarnström
Last updated: 2017-03-21