Kurtosis-Based Track Pruning for Sensor-Fusion Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Sensor-fusion systems face computational challenges in simultaneously tracking multiple classes of objects, such as vehicles and pedestrians, due to combinatorial explosions, which can lead to processing lag and inefficiencies.
Innovation Solution
The implementation of Kurtosis-based pruning, which determines matching errors between candidate object-tracks from different sensors, computes weights, and calculates a Kurtosis value to prune less relevant object-tracks, reducing the number of comparisons and preventing combinatorial explosions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor-fusion systems track multiple classes of objects simultaneously, then tracking accuracy and safety are improved, but computational complexity increases leading to combinatorial explosion
Solution Approach 1:
The patent segments the candidate object-tracks into different groups based on their Kurtosis values. By dividing the tracks into high-Kurtosis and low-Kurtosis groups, the system processes only the most relevant candidates in detail while efficiently discarding unlikely candidates, thus reducing computational complexity while maintaining tracking accuracy for multiple object classes
Solution Approach 2:
The patent introduces Kurtosis as a new parameter to characterize the distribution of matching errors. By computing and utilizing the Kurtosis value of matching error distributions, the system gains an additional dimension for filtering and pruning candidate tracks, enabling efficient reduction of computational complexity without sacrificing reliability
2Measurement precision
If sensor-fusion systems perform exhaustive matching of candidate object-tracks, then matching accuracy is improved, but processing speed decreases due to combinatorial explosion
Solution Approach 1:
The patent performs preliminary computation of Kurtosis values for all candidate object-tracks before executing the full matching process. This preliminary action identifies and prunes unlikely candidates in advance, so that the subsequent exhaustive matching is performed only on a reduced set of high-probability candidates, thereby maintaining matching accuracy while improving processing speed
Solution Approach 2:
The patent discards low-Kurtosis candidate tracks that are unlikely to be correct matches, reducing the computational burden. The pruning decision is based on the statistical property (Kurtosis) that captures the essence of matching quality, allowing the system to safely discard irrelevant candidates while preserving those that could lead to accurate matches
3Productivity
If sensor-fusion systems reduce the number of object-track comparisons, then processing efficiency is improved, but risk of discarding valid tracks increases
Solution Approach 1:
The patent uses Kurtosis, a statistical parameter that measures the tailedness of a distribution, to characterize matching error distributions. High Kurtosis values indicate that the matching errors are concentrated with few outliers, suggesting high confidence in the match. By thresholding on Kurtosis, the system prunes only low-confidence candidates while preserving high-confidence matches, thus improving processing efficiency without significantly compromising tracking reliability
Solution Approach 2:
The patent replaces the mechanical approach of exhaustive comparison with a statistical filtering mechanism based on Kurtosis analysis. Instead of blindly comparing all candidate pairs, the system uses the statistical property of matching error distributions to intelligently prune candidates, substituting computational brute-force with statistical reasoning to achieve both efficiency and reliability
Data Source
Figure 1
Figure 2
Figure 3-1
AI summary
This document describes Kurtosis based pruning for sensor-fusion systems. Kurtosis based pruning minimizes a total quantity of comparisons performed when fusing together large sets of data. Multiple candidate radar tracks may possibly align with one of multiple candidate visual tracks. For each candidate vision track, a weight or other evidence of matching is assigned to each candidate radar track. An inverse of matching errors between each candidate vision and each candidate radar track contributes to this evidence, which may be normalized to produce, for each candidate vision track, a distribution associated with all candidate radar tracks. A Kurtosis or shape of this distribution is calculated. Based on the Kurtosis values, some candidate radar tracks are selected for matching and other remaining candidate radar tracks are pruned. The Kurtosis aids in determining how many candidates to retain and how many to prune. In this way, Kurtosis based pruning can prevent combinatorial explosions due to large-scale matching.