RADAR Object Tracking with RLS Velocity Estimation in Clutter
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Extended Object Tracking (EOT) solutions face challenges in grouping return points from the same object together, associating new return points with existing tracks, updating tracks, and detecting noise in RADAR sensor data from automotive applications.
Innovation Solution
The proposed solution involves processing RADAR return points using density-based spatial clustering, Recursive Least Squares filters, and Kalman Filters to accurately compute the position, velocity, length, and width of objects, while filtering out noise and associating clusters with existing tracks, and initializing new tracks based on cluster characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If density-based spatial clustering is used to group RADAR return points, then object detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational process into distinct phases: return point filtering based on distance and motion criteria, density-based spatial clustering of valid points, and track association of clusters. This segmentation allows complex EOT to be broken into manageable steps, improving accuracy while controlling computational load at each stage.
Solution Approach 2:
The patent performs preliminary filtering of return points before clustering by removing points beyond threshold distance from map features and points with insufficient ergo-motion compensated range rates. This preliminary action reduces the number of points requiring complex clustering computation, thereby reducing overall computational complexity while maintaining detection accuracy.
2Measurement precision
If multiple RADAR time-steps are accumulated to increase point density, then tracking accuracy is improved, but processing time increases
Solution Approach 1:
The patent accumulates return points from multiple periodic RADAR time-steps to increase point density for clustering. By processing returns from several sequential time steps, the system improves tracking accuracy through better object shape and motion estimation while managing processing throughput.
Solution Approach 2:
The patent maintains continuous tracking by associating clusters from current time-step with existing tracks and updating track states. This continuity allows the system to leverage historical track information, reducing the need for complete re-processing and thereby managing processing time while maintaining accuracy.
3Productivity
If Linear Kalman Filters are used for velocity estimation, then computation efficiency is improved, but handling of non-linear motion dynamics is reduced
Solution Approach 1:
The patent introduces an intermediary velocity estimation step that computes velocities from cluster positions before feeding them to the Linear Kalman Filter. This intermediary computation transforms the problem into a form suitable for linear filtering, improving computation efficiency while the filter updates track states based on these estimated velocities.
Solution Approach 2:
The patent changes the state parameters of the track to include position, velocity, length, and width, and uses Linear Kalman Filters to update these parameters efficiently. By carefully selecting and updating only the necessary parameters, the system achieves computation efficiency while adequately representing object motion dynamics.
4Reliability
If return points are filtered by distance from map features and motion criteria, then noise rejection is improved, but false negatives increase
Solution Approach 1:
The patent applies different filtering criteria to return points based on their local characteristics: distance from prior map features and ergo-motion compensated range rates. By applying localized quality filters adapted to the specific context of each return point, the system improves noise rejection while minimizing false negatives through context-aware filtering.
Data Source
AI summary
Techniques are provided for extended object tracking using RADAR return points only. The techniques include receiving the return points from at least one RADAR sensor of the vehicle. One or more clusters based on or from the return points are generated. An estimated position and velocity of each of the one or more clusters is computed. A Recursive Least Squares (RLS) based algorithm is proposed to estimate the instantaneous velocity of a cluster in real-time that allows for accurate track-cluster association, removes the need to perform computationally expensive non-linear state updates, and allows for the estimation of the true velocity even in frames with a large amount of clutter. If it is determined that the one or more clusters are associated with an existing object track, the existing object track is updated using at least the respective positions of the one or more clusters associated with the existing object track.