NOTE: This repository is related with the next scientific work:
Barral, V.; Escudero, C.J.; García-Naya, J.A.; Maneiro-Catoira, R. NLOS Identification and Mitigation Using Low-Cost UWB Devices. Sensors 2019, 19, 3464.https://doi.org/10.3390/s19163464
If you use this code for your scientific activities, a citation is appreciated.
This node implements a positioning algorithm based on EKF to generate positions based on data coming from one or various sensors:
- Ranging values (distance between a tag and some reference anchors).
- Magnetic field values.
- Acceleration and gyroscope values.
- Rotation and displacement from an optical sensor.
To launch the node kfpos:
$ roslaunch gtec_roskfpos kfpos.launch Also there is another version that can be configured from command line to activate or deactivate the use of one or various sensors:
$ roslaunch gtec_roskfpos kfpos_with_args.launch This launcher file admits the following arguments:
- use_uwb Value of 1 to use the ranging measurements, 0 to ignore them.
- use_imu Value of 1 to use the IMU measurements, 0 to ignore them.
- use_mag Value of 1 to use the magnetometer measurements, 0 to ignore them.
- use_px4 Value of 1 to use the Px4Flow measurements, 0 to ignore them.
The node can be configured with the ROS topics that provide the sensors data. To do that, the next file must be edited:
catkin_ws/src/gtec/roskfpos/src/kfpos/config/config_subscriptions.xml
The file contains the name of the topics that publish the sensor data. It looks like:
<config>
<ranging use="1" topic="/gtec/toa/ranging"/>
<erlemag use="1" topic="/gtec/gazebo/erle/mag" interference="1" topicInterference="/gtec/gazebo/erle/maginterfered"/>
<erleimu use="1" topic="/gtec/gazebo/erle/imu"/>
<px4flow use="1" topic="/gtec/gazebo/px4flow"/>
<anchors use="1" topic="/gtec/gazebo/anchors"/>
</config>Value use="1" of each parameter indicates if the location algorithm must subscribe or not to this source of information.
Note: if ranging sensor is used, is also mandatory to be subscribed to the topic anchors, as this one publish the anchor positions and this information is needed by the location algorithm to process the ranging values.
The kfpos node can locate the objects in 2D or 3D. If the object is for example a vehicle, and its movement in Z axis is always zero, then the algorithm can be configured with a fixed value of height in order to improve the accuracy in the 2D estimation. To set the height of the ranging tag the next file should be edited:
catkin_ws/src/gtec/roskfpos/src/kfpos/config/config_t0.xml
Contents of this file look like:
<config>
<uwb useFixedHeight="1" fixedHeight="1.995"/>
</config>If parameter useFixedHeight="1" is set to 1, then the value set in fixedHeight (in meters) will be used to perform a 2D location.
The implemented algorithm has some internal options that can be used. The options are located in the following file:
catkin_ws/src/gtec/roskfpos/src/kfpos/config/config_pos.xml
This is the content of the file with a description of each option:
<config>
<!--
Algorithm
*********
type= 0: ML
type= 1: EKF
Variant: Algorithm variants
**************************************************
Variant= 0: Normal mode.
Variant= 1: The N rangings con more error are ignored.
- numIgnoredRangings Number of rangings to ignore.
Variant= 2: Only the best 3 anchors are used. - bestMode: Criterion to select the best anchors:
* 0: less error in x,y,z.
* 1: less error in z.
minZ, maxZ : Limits of Z coordinate.
***************************************
minZ: min Z value. Position is not generated if the estimation is lower.
maxZ: max Z value. Position is not generated if the estimation is greater.
EKF Initial position
****************
useInitPosition = 0: A random initial position is used to initialize the algorithm.
useInitPosition = 1: A fixed position is used. Position values are set using the next parameters:
- initX
- initY
- initZ
-->
<algorithm type="0" variant="0" numIgnoredRangings="2" bestMode="0" minZ="0.0" maxZ="3.0" useInitPosition="0" initX="8" initY="0" initZ="2"/>
</config>If a IMU is used, there are some configuration values that can be modify inside the next file:
catkin_ws/src/gtec/roskfpos/src/kfpos/config/config_erleimu.xml
Contents of file look like:
<config>
<erleimu useFixedCovarianceAcceleration="1" covarianceAcceleration="0.01" useFixedCovarianceAngularVelocityZ="1" covarianceAngularVelocityZ="0.000786"/>
</config>Parameters useFixedCovarianceAcceleration and useFixedCovarianceAngularVelocityZ indicate if a fixed value of covariance for the linear acceleration and for the angular velocity must be used. These fixed values and later set using the parameters covarianceAcceleration and covarianceAngularVelocityZ.
As in the previous case, there are some configuration parameters related with the magnetometer sensors that can be edited in the file:
catkin_ws/src/gtec/roskfpos/src/kfpos/config/config_erlemag.xml
File looks like:
<config>
<erlemag angleOffset="1.31559699349" covarianceMag="0.001"/>
</config>Parameter angleOffset can be used to set the sensors angle (in radians) between the sensor and the real north. Parameter covarianceMag can be used to set the covariance error of the sensor.
Px4Flow options can be configured using the next file:
catkin_ws/src/gtec/roskfpos/src/kfpos/config/config_px4flow.xml
File looks like:
<config>
<px4flow armP0="0" armP1="0.493" useFixedSensorHeight="1" sensorHeight="3.0" sensorInitAngle ="-1.570796326794897" covarianceVelocity="0.4" covarianceGyroZ="0.4"/>
</config>Parameters armP0 y armP1 can be used to select the coordinates of the arm where the Px4Flow sensor is placed on the side of the vehicle. Using useFixedSensorHeight parameter a fixed value of height can be used if the sensors is placed always at the same height. Parameter sensorHeight define this height.
Parameter sensorInitAngle (in radians) sets the initial angle between the vehicles orientation and the sensor. For example, an angle of 0 rads means that the vehicle is placed with the front looking at the positive X axis, whereas a -pi/2 rads angle would mean that the fronts vehicle is looking at the negative Y axis.
The last options, covarianceVelocity and covarianceGyroZ, can be used to set the covariance error of the velocity and gyro around Z axis.