Object Tracking Device Grouping for Reduced Computational Load

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing object tracking systems face increased workload when processing a large number of objects in a time series, particularly in pedestrian detection and driving support devices, which can lead to inefficiencies and potential safety issues.

Innovation Solution

An object tracking method that groups detected objects based on their influence on the movement of a host vehicle, allowing each group to be tracked as a single object in a time series, thereby reducing the computational load.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If all detected objects are tracked individually in a time series, then tracking precision is improved, but calculation workload increases

Engineering Contradiction:
Improvetracking precisionVSAvoidcalculation workload
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges multiple detected objects into a single tracking target when they satisfy specific conditions: same detection box ID, distance between centers less than threshold, and similar movement directions. This combining approach reduces the number of individual tracking operations while maintaining accurate tracking of the merged group, thereby decreasing calculation workload without significantly compromising tracking precision.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent segments the tracking process into two distinct phases: object detection phase and object tracking phase. In the detection phase, multiple objects are identified individually with high precision. In the tracking phase, these detected objects are evaluated against merging conditions, and only those that meet the criteria are combined. This segmentation allows the system to maintain detection precision while reducing tracking complexity through selective merging.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If all detected objects are tracked individually in a time series, then tracking accuracy is improved, but processing time increases

Engineering Contradiction:
Improvetracking accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

By merging multiple objects that move together and satisfy the specified conditions into a single tracking target, the patent reduces the total number of tracking operations required. This directly decreases processing time while maintaining tracking accuracy for the merged group, as the system tracks the collective movement rather than individually processing each object within the group.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent applies partial merging rather than merging all detected objects. Only objects that satisfy the merging conditions (same box ID, close distance, similar direction) are combined, while others remain as individual tracking targets. This selective partial action optimizes processing time by reducing redundant calculations only where appropriate, without sacrificing tracking accuracy for objects that should be tracked separately.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3477615B1Object tracking method and object tracking device
Publication Date: 2022.03.09 NISSAN MOTOR CO LTD
  • EP3477615B1 patent drawingFigure 1
  • EP3477615B1 patent drawingFigure 2
  • EP3477615B1 patent drawingFigure 3

AI summary

An object tracking device tracks objects (8) in a time series using an object detection sensor (10) for detecting objects (8) around a host vehicle (7). When a plurality of objects (8) are detected, the object tracking device sorts the objects (8) into groups (G) in accordance with a degree of influence of the positions of the objects (8) on the movement (9) of the host vehicle (7). The object tracking device tracks each of the groups (G) as a single object in a time series.