Simple odometry optimizer ROS node

Initial setup

Download your rosbag and move it to src/odometry_optimizer/rosbag/odometries.bag

Build and run

First, ensure catkin tools is installed (so you can catkin build)

sudo apt install ros-melodic-catkin python-catkin-tools

Then initialize the workspace and build

catkin init
catkin build --cmake-args -DCMAKE_BUILD_TYPE=Release

Optional: Set up rovio as VO frontend

First, set up kindr, as explained here: https://github.com/ethz-asl/kindr.

Then do the following:

cd src
git clone [email protected]:ethz-asl/rovio.git
cd rovio
git submodule update --init --recursive # gets lightweight_filtering
cd ../..
catkin build --cmake-args -DCMAKE_BUILD_TYPE=Release

Run with rovio

roslaunch odometry_optimizer rovio_solution.launch

Topics

The node subscribes to odometries from the topic /rovio/odometry and publishes the optimized path on /optimized_pose. You can see the outgoing poses with the command

rostopic echo /optimized_pose

Rviz

You can view the optimized path in rviz by adding a path visualization on the /optimized_pose topic.

Export bag of optimized poses

To export a bag of optimized poses to rosbag/paths.bag, run

roslaunch odometry_optimizer record.launch

Simulating failure cases

You can simulate failure of sensors by disabling the incoming sensor streams. Both rovio and lidar can be disabled. To disable one of the streams, run any of the following commands:

rosservice call /toggle_rovio
rosservice call /toggle_lidar

As the service names imply, this toggles the streams, so to turn them on again simply run the command once more.

Running tests 2.0

Run tests with

catkin_make run_tests odometry_optimizer

To actually get a non-zero return code for failing tests, run the following command after running the tests

catkin_test_results

It will return 0 if all tests pass, and 1 otherwise.