Object Detection Apparatus Scoring Reliability

Solution Overview

Problem

Existing object detection apparatuses, such as those described in JP-B2-3903856, fail to accurately determine the presence of target objects due to reliance on individual observation timings, leading to uncertainty and potential delays or erroneous detections, as they do not consider past observation results when calculating reliability levels.

Innovation Solution

An object detection apparatus that includes an information acquiring unit, a target object recognizing unit, a predicting unit, a score deriving unit, and a reliability level deriving unit, which acquires sensor information, predicts the state of a target object at the next observation timing, calculates a score based on the difference between observed and predicted states, and statistically processes scores across multiple observation timings to derive a reliability level that indicates the certainty of the target object's presence.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If reliability level is calculated based on individual observation timing only, then calculation is simple, but detection accuracy deteriorates due to uncertainty and potential delays

Engineering Contradiction:

Improvedetection accuracyVSAvoidcalculation complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary prediction of the target object's state at the next observation timing based on current recognition information. This prediction is prepared in advance to compare with actual future observations, enabling proactive reliability assessment rather than reactive calculation after uncertainty arises.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system establishes a feedback loop where predicted states are continuously compared with actual observed states. The degree of difference between predicted and actual states feeds back into reliability level calculation, creating a self-correcting mechanism that improves detection accuracy over time by learning from observation discrepancies.

Inventive Principle:

Principle #23Feedback

2Reliability

If past observation results are not considered, then processing speed is fast, but reliability deteriorates due to inability to suppress uncertainty

Engineering Contradiction:

Improvedetection reliabilityVSAvoidprocessing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary prediction of the target object's state at the next observation timing based on current recognition information. This prediction is prepared in advance to compare with actual future observations, enabling proactive reliability assessment rather than reactive calculation after uncertainty arises.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system establishes a feedback loop where predicted states are continuously compared with actual observed states. The degree of difference between predicted and actual states feeds back into reliability level calculation, creating a self-correcting mechanism that improves detection accuracy over time by learning from observation discrepancies.

Inventive Principle:

Principle #23Feedback

3Measurement precision

If statistical processing of scores across multiple observation timings is performed, then detection accuracy is improved, but computational complexity increases

Engineering Contradiction:

Improvedetection accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the reliability assessment into discrete scoring events at each observation timing. Instead of continuously processing all historical data, it calculates scores at specific observation points and statistically processes these segmented scores, reducing computational complexity while maintaining accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs statistical processing on a selective subset of observation scores rather than all possible historical data. By focusing on relevant observation timings and using predictive comparison, it achieves sufficient detection accuracy without the excessive computational burden of processing every available data point.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS11614536B2Object detection apparatus based on scoring reliability and increasing relative range rate

Publication Date: 2023.03.28 DENSO CORP

AI summary

An object detection apparatus detects a target object present in a periphery of a moving body. The object detection apparatus derives recognition information indicating a state of a target object, and predicts a state of the target object at a next second observation timing, based on the recognition information derived at a first observation timing. The object detection apparatus derives a score based on a degree of difference between a state of the target object observed at the second observation timing and a next state of the target object predicted at the first observation timing. The object detection apparatus derives a reliability level by statistically processing scores related to the target object derived at a plurality of observation timings from past to present. In response to the reliability level satisfying a predetermined reference, the object detection apparatus determines that the target object related to the reliability level is actually present.