Field Tests

Field Tests

Plan description

The first phase of the project is the acquisition of all navigation data (sensors, GPS, SERVO control) in order to trim the navigation algorithms with real data. The data will be used in order to simulate the vehicle and to verify the output of the navigation algorithm and INS filtering. In this part the analysis will be done using the Matlab software tool and we will check all the systems phases from the input IIR filters to the output SERVO commands.

First Step: Data Acquisition

After the software and hardware implementation has finished the next step is to test the whole system in a real scenario. First we will collect the full data in a predefined route in order to emulate the navigation algorithm and make all the necessary modifications to the system’s hardware and software in order to adapt it to the real “life”.
The first step is to collect and examine all the sensors and GPS pure data. With the first look in the pure data we can not extract enough information as we can see in the following graphs. So, we conclude to the obvious, signal processing is necessary.

con_field_clip_image002

con_field_clip_image003

con_field_clip_image004

Pure Sensor Data

The results are better when we apply the necessary filtering in the incoming data from the sensors. In the following graphs we can see the output data from the low pass filters that will feed the INS. With a short view in the following graphs, we can notice the noise reduction after filtering. Also, we can easily notice the correlation between the Gyro’s outputs the compass’s outputs, this will be very helpful to improve the performance of the INS when running along without any GPS data. Note that the three rate gyro sensors and the three compass sensors feed the INS with almost the same information (heading information) and also if the vehicle is an aircraft then data coming from compass sensor are very accurate since no external magnetic fields can easily affect the performance of the sensor. Heading information is very critical in the overall system performance when the GPS data absent. This information will be used to reject the earth gravity from the accelerometer outputs.
Finally, we can see the two small section with no GPS data.

Field Data Part I

con_field_clip_image005
con_field_clip_image006
con_field_clip_image007
INS Data (Acceleration, Rate Gyro, Compass)

con_field_clip_image008
con_field_clip_image009
con_field_clip_image010

GPS Data (Velocity, No of Satellites)

con_field_clip_image011

GPS Position

Now, we can focus on a smaller part of samples, about 45sec, to test the INS algorithm.

First of all, we must integrate the gyro output in order to have the rotation of the vehicle from the initial position. The sensor output voltage changes by 5mV for each degree of rotation per second. So, we can change the Y axis to degrees from Volts. In the following graph we can see the data from the gyros. In the Yaw (X axis) we have a grate change, about 85 degrees, and also we can notice an increase for the Pitch (Y axis).

con_field_clip_image012
Gyro Output (After Processing)

We can process the compass output for the same duration in order to compare it with the results from the gyros.
The earth magnetic field gives us a variation of ±25mV on the compass sensor output. So, we can calculate the heading information of the vehicle using the following equation:

con_field_clip_image015

Compass Roll Output

It is time to examine the output of the acceleration sensors. The first step is to integrate the output. Before this it is necessary to remove the offset of the sensor and then make a correction for the earth gravity. For this data examination will use the previous calculated rotation from the rate gyro outputs as reference. The following graph presents the acceleration of the vehicle after offset and earth gravity removing:

con_field_clip_image016

Vehicle Acceleration

The next step is to integrate the ‘clean’ acceleration in order to obtain the vehicle velocity for the three axes. We can see the vehicle velocity in the following graph:

con_field_clip_image017
Vehicle Velocity (INS Only)

Now, we can integrate one more time the velocity in order to get the displacement of the vehicle in the three axes:

con_field_clip_image018

Position Displacement (INS Only)

The above graph will help us to calculate the vehicle route. In order to calculate the route we need information for the rotation of the vehicle.  For the route calculation we will use the output of the rate gyros. We will calculate the position using the following equation:

con_field_clip_image023

INS Route

From the atmospheric pressure we can calculate the altimeter of the vehicle. The output of the pressure sensor increases 45.9mV for each kPa of pressure change. The transfer function of the sensor is:

uav   (Volt)
Also,    uav       (feet)

The following graph shows the altitude, calculated from the pressure sensor:

uav

Altimeter

Finally, we must mention the temperature changes inside the sensors that affect the offset voltage of the sensors output. In the following graph we can see the changes in the temperature from an embedded temperature sensor inside to the gyroscope. Correction needed in order to reduce the effects of those changes to the INS.

con_field_clip_image029
Temperature

Comparing the results from the INS with those from the GPS we can see that the output from the INS is extremely good. Note, that in the calculations we have discarded many parameters that will cause improvement to the INS output (Temperature compensation, better earth gravity correction, better sensor offset calculations and many others). From the first field test we can see that the vehicle can be self guided without GPS information for many seconds (in the last analysis the performance seems absolutely good for the duration of 45sec that examined).