Integration RSSI Localization

[reinzor]

Hi,

I was thinking of integrating RSSI measurements (indoor instead of GPS) with this package; does anybody have experience with this? I've found some papers online and these approaches seem to be accurate up to ~1 meter.

I see now some options for the integration

• Fusing RSSI measurements into a sensor_msgs/NavSatFix and convert it using the navsat_transform_node to odom
• Fusing RSSI measurement to nav_msgs/Odometry directly
• Integrating RSSI measurements into the filter..

Let me know if you have any experience or thoughts.

[AlexisTM]

@reinzor I am going to use a UWB localization and I will be in the same position as you soon. I think you send your data as an Odometry message and sending a covariance of 99999 for elements not computed.

[AlexisTM]

@reinzor Check that video! Awesome information :D

http://www.ros.org/news/2016/06/tom-moore-working-with-the-robot-localization-package.html

[reinzor]

Hi @AlexisTM , thanks I know that video :). But looking back at this, I am wondering how to calibrate the RSSI position odometry with the base odometry. In the navsat_transform node this is explicitly done so that the gps odom and wheel odom have the same starting point.

[eroniki]

@reinzor I have implemented a similar localization scheme. My RSSI localizer provides a global location, not the pose, by using trilateration method. In my case, the resulting estimations are converted PoseWithCovarianceStamped messages. Therefore, the robot_localization package was able to use the estimations as sensor measurements. It worked well for my case and I would recommed using in the same scheme.

[reinzor]

Hi Murat, thanks for the information! We are planning the same approach.

-Rein

On Thu, Jan 25, 2018 at 3:37 PM, Murat Ambarkutuk notifications@github.com wrote:

@reinzor https://github.com/reinzor I have implemented a similar localization scheme. My RSSI localizer provides a global location, not the pose, by using trilateration method. In my case, the resulting estimations are converted PoseWithCovarianceStamped http://docs.ros.org/api/geometry_msgs/html/msg/PoseWithCovarianceStamped.html messages. Therefore, the robot_localization package was able to use the estimations as sensor measurements. It worked well for my case and I would recommed using in the same scheme.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cra-ros-pkg/robot_localization/issues/352#issuecomment-360484633, or mute the thread https://github.com/notifications/unsubscribe-auth/AD-4l7ZlOHoEli9XBi2DpcFyzz46kYxQks5tOJGbgaJpZM4MT1Ct .