Navigation on Fixed Paths

Most mobile robots navigate along fixed paths. LiDAR-localization generally is the standard for localization as it provides good performance with low commissioning effort. However, LiDAR-localization performance can suffer in dynamic environments and might require infrastructure such as reflectors, codes or tape to perform well. This adds commissioning complexity, time and costs.

Motivating example: an application with conveyor mobile robots driving on fixed paths (shown in red and blue). Pick-up and drop-off points are marked by circles. Two charging stations are present, in which the vehicle needs to drive with high accuracy (paths shown in yellow).

The major benefits of combining LiDAR-localization and Triton Floor-LOC are:

• LiDAR-localization enables you to localize with respect to a globally accurate map
• Triton Floor-LOC keeps the accuracy from day 1, even in dynamic environments
• Triton Floor-LOC offers sub-millimeter level repeat accuracy, ideal at pick-up and drop-off points
• Both require no infrastructure

This article presents an efficient workflow for commissioning a project using Triton Floor-LOC and LiDAR-localization.

In short, the workflow can be summarized as follows:

1. Create a LiDAR map
   Explore the area by driving a robot around while the LiDAR-localization software builds a map of the environment using its laser-scanner and odometry
2. Plan main navigation paths
   Define the main paths where the vehicles will drive and points where the vehicle should stop in your navigation software
3. Create Triton map on main navigation paths
   Let your vehicle drive autonomously along the main navigation paths while Triton’s mapping mode is enabled.
4. Optional: Teach & repeat docking
   To navigate with millimeter precision into a docking station, manually drive in to/out of the docking station to create the Triton map exactly where you need it.

To be able to follow this guide successfully, we make the following assumptions:

• Your LiDAR-localization performs well, with an accuracy of about 2cm
• You have a navigation software where you can define navigation paths and points of interest (pick-up/drop-off points, docking stations)

2.1.1. Create LiDAR map

Info: For this guide we will be using SICK’s LiDAR-LOC 2.10 for LiDAR-localization. The map is created using the Mobile Robot Manager extension in Sentio Creator and also allows to add navigation paths and nodes. Click here to learn how to use the Mobile Robot Manager.

Procedure:

• Explore the area by driving a robot around (follow the best practices/instructions from your LiDAR-localization software):

Figure: Mobile Robot Manager is used to create a LiDAR-LOC map while driving around.

• Result: The robot built a map of the environment using its laser-scanner and odometry:

Figure: LiDAR-LOC map is finalized in the Mobile Robot Manager software.

• The LiDAR map can now be used for localization

2.1.2. Plan virtual lines

• Draw curves and straight line sections in your navigation software
• Add nodes at points of interest (drop-off/pick-up points, docking stations, …)

Figure: Navigation paths and nodes been added to the LiDAR-LOC map in Mobile Robot Manager.

2.1.3. Create Triton map

• Ensure your vehicle is localized based on LiDAR-localization
• Provide external reference to Triton
• • If using LiDAR-LOC 2.10: this is done automatically when you configured Triton
  • For other users (ROS/ROS2/C++): Use the setup you created in the software integration guide procedure: Create a Triton map - Procedure: Send external reference to Triton
• Position vehicle at first node
• Enable mapping mode on Triton
• Let vehicle drive along navigation paths
• When vehicle returns to initial node: disable mapping mode
• Repeat until you covered all navigation paths while Triton was mapping.

Warning: Try to avoid clusters overlapping more than 10cm. Some overlap is desired (in case you want to increase the accuracy of your map later using Map Optimization), but excessive overlap will reduce localization performance.

Video: Example mapping of navigation paths. Mapping mode is started/stopped manually via SOPAS-Air to have some overlap (playback speed x2). Floor map is visualized by purple circles (do not exactly align with the navigation path as the Triton is in front of the robot).

2.1.4. Optional: Teach and Repeat Triton mapping for docking stations

For our example mobile robot use-case from Figure 1 we assume that there are charging stations that are fixed in place. Driving your vehicle autonomously into such a docking station based on LiDAR localization might not be feasible:

• it is difficult to draw virtual lines that go exactly into the docking station
• it is difficult to determine the exact end position of where the robot should stop
• the repeat accuracy of your LiDAR-localization might be insufficient to drive autonomously into the docking station

For these cases we recommend to create the Triton map, and determine the path and end position “Teach and repeat” style. Please read our application document on Teach & Repeat Docking in the section below with related articles.

In case the charging station would not be fixed to the floor, we could have used the same procedure as in 2.1.3., let the vehicle drive autonomously into the docking station and adjust the position of the docking station according to the end position of the vehicle.

Video: To get the best accuracy possible, we mapped the docking station “Teach and Repeat” style.

Info: After following the procedures above you now have two sources of localization available (LiDAR-based localization and Triton). It is up to the user whether you want to switch between sources (use Triton when available, otherwise switch back to LiDAR), fuse the two based on uncertainties or just use Triton.

2.2.1. Localization

• Update the pose of your robot based on LiDAR-localization and/or Triton corrections
• Provide external reference to Triton (best estimate of Triton’s pose)

Note: LiDAR-LOC 2.10 will automatically include Triton corrections in its localization and send the external reference to Triton.

2.3.1. Adjust navigation paths (if needed)

• Re-use the virtual lines from section 2.1.2. Plan virtual lines
• If you mapped the docking station Teach & Repeat style: Add/adjust virtual lines and end positions based on the values you wrote down during mapping. Make sure the virtual line ends exactly where the robot should end.

Video: A navigation path is added based on the begin and end pose from the teach & repeat docking procedure. This docking path is then connected to the main navigation paths.

• Drive vehicle across all lines and points of interest to test localization

Video: Vehicle drives navigation paths while Triton corrections are combined with LiDAR-localization (LiDAR-LOC). At the end of the video the vehicle drives into the docking stations and comes to a stop.

In this application note we have showed how to:

• Create a contour map (using LiDAR-LOC and Mobile Robot Manager)
• Draw navigation paths and map those using Triton Floor-LOC
• Commission a docking station 'Teach & repeat'-style

We recommend to explore the following articles:

• Analyze Triton floor maps and logs: Learn how to use our Python tooling to analyze the Triton floor map you created and localization performance using Triton logs.