On-demand vehicle lost item detection using feature vector comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing on-demand vehicle management systems face challenges in accurately detecting lost items due to incorrect identification of objects and users, particularly in shared vehicle services where multiple users interact at different locations.
Innovation Solution
An on-demand vehicle management device equipped with a communication unit, categorization information generation unit, and warning notification unit that associates detected objects with users based on sensor data, compares pre- and post-vehicle usage categorization information to determine changes in user possessions, and outputs warnings for accurate lost item detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the vehicle asks the user whether they have left anything in the vehicle when detecting a person or object, then the system can detect objects in the vehicle, but the detection accuracy deteriorates due to false positives from objects not belonging to the user or from other users remaining in the vehicle
Solution Approach 1:
The system performs preliminary actions by capturing images and extracting feature vectors before the user actually uses the vehicle. It stores these feature vectors in advance and compares them with current detection data to determine object ownership, thereby preventing false positives before they occur
Solution Approach 2:
The system uses feedback by continuously comparing current detection data with previously stored feature vectors and user information. This feedback mechanism allows the system to identify whether detected objects belong to the current user or are false positives, improving detection accuracy through iterative verification
2Reliability
If the system detects any object in the vehicle as a potential lost item, then it can alert users to check for lost items, but it generates false alarms when objects belong to other users or are not the current user's possession
Solution Approach 1:
The system performs preliminary actions by capturing images and extracting feature vectors before the user actually uses the vehicle. It stores these feature vectors in advance and compares them with current detection data to determine object ownership, thereby preventing false positives before they occur
Solution Approach 2:
The system uses an intermediary approach by introducing a feature vector extraction and comparison mechanism as a mediator between object detection and user identification. This intermediary layer filters out false positives by comparing detected object features with stored user-associated features before generating warnings
Data Source
AI summary
An on-demand vehicle management device includes a communication unit that receives detection data output by at least one of a sensor provided in or on an on-demand vehicle or a sensor installed along a travel route, a categorization information generation unit that recognizes a user, based on the detection data, and generates categorization information that associates an object represented by information included in the detection data with the user, as a possession of the user, a categorization information comparison unit that determines whether a change has occurred in a situation of the possession by making a comparison, on a per user basis, between the categorization information as of before the user gets on the on-demand vehicle and the categorization information as of when the user gets off the on-demand vehicle, and a warning notification unit that outputs warning information when it is determined that the change has occurred.