Abstract:A 2D range-only tracking scenario is non-trivial due to two main reasons. First, when the states to be estimated are in Cartesian coordinates, the uncertainty region is multi-modal. The second reason is that the probability density function of azimuth conditioned on range takes the form of a generalized von Mises distribution, which is hard to tackle. Even in the case of implementing a uni-modal Kalman filter, one needs expectations of trigonometric functions of conditional bearing density, which are not available in the current literature. We prove that the trigonometric moments (circular moments) of the azimuth density conditioned on range can be computed as an infinite series, which can be sufficiently approximated by relatively few terms in summation. The solution can also be generalized to any order of the moments.
This important result can provide an accurate depiction of the conditional azimuth density in 2D range-only tracking geometries. We also present a simple optimization problem that results in deterministic samples of conditional azimuth density from the knowledge of its circular moments leading to an accurate filtering solution. The results are shown in a two-dimensional simulation, where the range-only sensor platform maneuvers to make the system observable. The results prove that the method is feasible in such applications.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the main challenges encountered in 2D range - only tracking. Specifically, the author focuses on:
1. **Multimodal Uncertainty Region**: When the state is represented in Cartesian coordinates, the uncertainty region exhibits a multimodal distribution, which makes the use of the traditional Gaussian assumption invalid.
2. **Complexity of Conditional Azimuth Density**: Given the distance, the probability density function of the azimuth angle takes the form of a generalized von Mises distribution, which is difficult to handle. In particular, calculating the trigonometric moments (or circular moments) of the conditional azimuth density is crucial for achieving effective filter updates, but these moments are not available in the existing literature.
To solve these problems, the author proposes the following methods:
- **Deriving a General Solution for Conditional Integration**: The author derives the conditional integrals \(E[\cos(m\theta)|r]\) and \(E[\sin(m\theta)|r]\), where \(\theta\) is the azimuth angle, \(r\) is the distance, and \(m\) is a positive integer. These integrals can be expressed as infinite series and can be sufficiently approximated by a finite number of terms.
- **Deterministic Sampling**: Based on the above - derived trigonometric moments, the author proposes an optimization method to generate deterministic samples from the conditional azimuth density. These samples are used to construct more accurate posterior estimates.
- **Application to 2D Range Tracking Problems**: Through simulation verification, the feasibility of this method in practical application scenarios is demonstrated.
In summary, this paper aims to provide an effective method for calculating the trigonometric moments of the conditional azimuth density and improve the performance and accuracy of the 2D range - only tracking system through the deterministic sampling technique.
### Formula Summary
- Conditional Integration Formulas:
\[
E[\cos(m\theta)|r]=\int_{0}^{2\pi}\cos(m\theta)p(\theta|r)d\theta
\]
\[
E[\sin(m\theta)|r]=\int_{0}^{2\pi}\sin(m\theta)p(\theta|r)d\theta
\]
- True Range Density:
\[
p(r_{k})=\kappa(r_{k})2\pi\left[I_{0}(A_{3}r_{k}^{2})I_{0}(r_{k}\sqrt{A_{1}^{2}+A_{2}^{2}})+2\sum_{k = 1}^{\infty}I_{k}(A_{3}r_{k}^{2})I_{2k}(r_{k}\sqrt{A_{1}^{2}+A_{2}^{2}})\cos(2k\psi)\right]
\]
- Conditional Azimuth Density:
\[
p(\theta_{k}|r_{k})=\frac{\kappa_{k}}{p(r_{k})}\exp\left[r_{k}\|p,q\|\cos(\theta_{k}-\phi_{1})+A_{3}r_{k}^{2}\cos(2\theta_{k}+\phi_{2})\right]
\]
- General Expressions for Trigonometric Moments:
\[
I_{\text{gen},\cos(m)}=\frac{2\pi}{\Delta_{k}}\left[\cos\left(\frac{m\phi_{2}}{2}\right)I_{\text{diag},\cos}+\sin\left(\frac{m\phi_{2}}{2}\right)I_{\text{diag},\sin}\right]
\]
\[
I_{\text{gen},\sin(m)}=\frac{2\pi}{\Delta_{k}}\left[\cos\left(\frac{m\phi_{2}}{2}\right)I_{\text{diag},\sin}-\sin\left(\frac{m\phi_{2}}{2}\right)I_{\text{diag},\cos}\right]
\]