Hostname: page-component-7b7567c8c-wxvwp Total loading time: 0 Render date: 2026-11-17T06:58:08.019Z Has data issue: false hasContentIssue false

Classification of UAV and bird target in low-altitude airspace with surveillance radar data

Published online by Cambridge University Press:  14 March 2019

W. S. Chen*
Affiliation:
Airport Research Institute, China Academy of Civil Aviation Science and Technology, Beijing, China
J. Liu
Affiliation:
Airport Research Institute, China Academy of Civil Aviation Science and Technology, Beijing, China
J. Li
Affiliation:
Airport Research Institute, China Academy of Civil Aviation Science and Technology, Beijing, China
Rights & Permissions [Opens in a new window]

Abstract

In order to ensure low-altitude safety, a tracking and recognition method of unmanned aerial vehicle (UAV) and bird targets based on traditional surveillance radar data is proposed. First, several motion models for UAV and flying bird targets are established. Second, the target trajectories are filtered and smoothed with multiple motion models. Third, by calculating the time-domain variance of the model occurrence probability, the model conversion probability of the target is estimated, and then the target type is identified and classified. The effectiveness and robustness of the algorithm is demonstrated by several groups of Monte Carlo simulation experiments, including setting different recognition steps, different model transformation probability, filtering and smoothing algorithm comparison. The algorithm is also successfully applied on the ground-truth radar data collected by the low-altitude surveillance radar at airport and coastal environments, where the targets of UAVs and flying birds could be tracked and recognised.

Information

Type
Research Article
Copyright
© Royal Aeronautical Society 2019 
Figure 0

Figure 1Classification scheme for unmanned aerial vehicle (UAV) and flying bird targets.

Figure 1

Figure 2Simulated trajectories of UAV and flying bird targets.

Figure 2

Figure 3RMSE of target tracking.

Figure 3

Table 1RMSE of simulated target tracking by 1,000 Monte Carlo experiments

Figure 4

Figure 4Estimation of model probability in the tracking of UAV target.

Figure 5

Figure 5Estimation of model probability in the tracking of flying bird target.

Figure 6

Figure 6ROC curves with the filtering algorithm by different number of recognition steps.

Figure 7

Figure 7ROC curves with the filtering algorithm under different probabilities of model change.

Figure 8

Figure 8ROC curves with the smoothing algorithm under different probabilities of model change.

Figure 9

Figure 9Comparison of the ROC curves between the filtered and smoothed algorithm.

Figure 10

Figure 10Surveillance radar at airport.

Figure 11

Figure 11Radar surveillance data at airport.

Figure 12

Figure 12Classification of UAV and bird targets with case 1.

Figure 13

Table 2Recognition results of UAV and bird targets for case 1

Figure 14

Figure 13Surveillance radar at coastal area.

Figure 15

Figure 14Classification of UAV and bird targets with case 2.

Figure 16

Table 3Recognition results of UAV and bird targets for case 2