Linear Recursive Automotive Target Tracking Filter for Advanced Collision Warning Systems

Abstract

This paper proposes an improved automotive target tracking scheme using FMCW radar which is necessary for the advanced collision warning systems. Since there exist strong nonlinear relationships between the FMCW radar measurements and the target state, the target tracking and data association in dense road clutters have been recognized as a quite challenging problem. It is obvious that the use of accurate range rate measurement might be an excellent choice to improve both target tracking and clutter suppression performances. This motivates us to develop a novel linear recursive automotive target tracking filter based on the measurement conversion in the predicted line-of-sight (LOS) Cartesian coordinate system (PLCCS). Since the x axis of the PLCCS is set by the predicted LOS vector from the host to the target, if the LOS prediction error is imperceptible, the range rate can be approximated to the x component of the relative target velocity in PLCCS. Employing the PLCCS drastically reduces the complexity of the problem and allows us to solve it within the framework of linear recursive Kalman filtering. Through the simulations, the superiority of the proposed method is compared to the existing nonlinear automotive target tracking filters.

Hybrid Interacting Multiple Model Filtering for Improving the Reliability of Radar-Based Forward Collision Warning Systems

“…The sampling interval, target detection probability, gate probability, and gate threshold were set as s, , , and , respectively. The noise variances of the range/bearing measurements were set as m and [ 9 , 10 ]. Trajectories of preceding vehicle were generated using the DWNA model ( 1 ) with the standard deviation of process noise, .…”

“…Studies on FCW algorithms focused on the decision-making process involved in determining whether to alert a collision. In typical FCW systems [ 9 , 10 , 11 , 12 ], the time-to-collision (TTC) is computed using the information of the relative distance and velocity between the host and preceding vehicles, and a warning is provided if the TTC was below a certain threshold. Recent studies [ 13 , 14 , 15 ] analyzed the behaviors (driving patterns) of vehicles using artificial intelligence (AI) algorithms, and a warning is provided in situations with a high risk of collision.…”

Hybrid PDA/FIR Filtering for Preceding Vehicle Tracking Using Automotive Radars

“…Simulations were conducted for the eight motions described above, two clutter conditions (moderate and heavy clutters), and six detection probabilities. The preceding vehicle tracking was conducted for 19.5 s. The sampling interval was set as T = 0.3 s. The automotive radar reported range/bearing measurements with noise covariances σ r = 0.25 m and σ theta = 1.5 • [20], [24]. The clutter density λ was set to 1.0 × 10 −2 and 1.0 × 10 −1 for the moderate and heavy clutter environments, respectively.…”

“…The measurement validation process selects the measurements within the validation region (also called the gate) as follows [5], [6], [20], [21]:…”

Gaussian mixture approach to decision making for automotive collision warning systems

“…To identify the correct target measurement among multiple radar measurements from the cluttered environment, various data association techniques can be applied for designing the automotive target tracking filter [21][22][23]. In this paper, the target tracking filter using pseudo-measurements defined in line-of-sight Cartesian coordinate is used for estimating the states of automotive targets [24]. Using the target available information, the threat measure generation unit estimates the TTC and predicts the collision position of a threat vehicle.…”