Post-Processing Kalman Filter Application for Improving Cooperative Awareness Messages’ Position Data Accuracy
Cooperative intelligent transportation systems continuously send self-referenced data about their current status in the Cooperative Awareness Message (CAM). Each CAM contains the current position of the vehicle based on GPS accuracy, which can have inaccuracies in the meter range. However, a high ac...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-12-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/24/24/7892 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Cooperative intelligent transportation systems continuously send self-referenced data about their current status in the Cooperative Awareness Message (CAM). Each CAM contains the current position of the vehicle based on GPS accuracy, which can have inaccuracies in the meter range. However, a high accuracy of the position data is crucial for many applications, such as electronic toll collection or the reconstruction of traffic accidents. Kalman filters are already frequently used today to increase the accuracy of position data. The problem with applying the Kalman filter to the position data within the Cooperative Awareness Message is the low temporal resolution (max. <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>10</mn><mo> </mo><mi mathvariant="normal">H</mi><mi mathvariant="normal">z</mi></mrow></semantics></math></inline-formula>) and the non-equidistant time steps between the messages. In addition, the filter can only be applied to the data retrospectively. To solve these problems, an Extended Kalman Filter and an Unscented Kalman Filter were designed and investigated in this work. The Kalman filters were implemented with two kinematic models. Subsequently, driving tests were conducted with two V2X vehicles to investigate and compare the influence on the accuracy of the position data. To address the problem of non-equidistant time steps, an iterative adjustment of the Process Noise Covariance Matrix <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi mathvariant="bold-italic">Q</mi></mrow></semantics></math></inline-formula> and the introduction of additional interpolation points to equidistance the received messages were investigated. The results show that without one of these approaches, it is impossible to design a generally valid filter to improve the position accuracy of the CAM position data retrospectively. The introduction of interpolation points did not lead to a significant improvement in the results. With the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi mathvariant="bold-italic">Q</mi></mrow></semantics></math></inline-formula> matrix adaptation, an Unscented Kalman Filter could be created that improves the longitudinal position accuracy of the two vehicles under investigation by up to 80% (0.54 m) and the lateral position accuracy by up to 72% (0.18 m). The work thus contributes to improving the positioning accuracy of CAM data for applications that receive only these data retrospectively. |
|---|---|
| ISSN: | 1424-8220 |