Elevator Boarding Prediction Using Passenger Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Elevator congestion is exacerbated by passengers not getting into the elevator despite it stopping at their floor, due to factors like passenger features, car occupancy, and cultural context, which existing systems fail to predict accurately.
Innovation Solution
An elevator system that uses image acquisition, learning data generation, and machine learning to estimate whether a passenger will board, thereby optimizing elevator operations based on real-time passenger features and situational data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the elevator stops at every calling floor, then all waiting passengers can potentially board, but unnecessary stops increase and transport efficiency decreases
Solution Approach 1:
The system performs preliminary action by capturing images of waiting passengers before the elevator arrives, analyzing their features (gender, height, age group) in advance, and predicting whether they will board. This allows the elevator control system to make informed decisions about whether to stop at each floor before actually arriving, avoiding unnecessary stops and improving transport efficiency.
2Productivity
If the elevator skips floors based on simple occupancy rules, then unnecessary stops are reduced, but prediction accuracy decreases due to ignoring passenger features and cultural context
Solution Approach 1:
The system applies local quality by analyzing specific features of waiting passengers (gender, height, age group) and the elevator car's current state (occupancy, passenger composition) to make localized predictions for each floor. This granular analysis considers cultural context and interpersonal dynamics, significantly improving prediction accuracy compared to simple occupancy-based rules.
3Measurement precision
If the system collects and analyzes detailed passenger features, then prediction accuracy improves, but system complexity and data processing requirements increase
Solution Approach 1:
The system replaces complex mechanical or manual analysis with automated image recognition and machine learning algorithms. Cameras capture passenger images, which are then automatically processed by AI models to extract features (gender, height, age group) and predict boarding behavior. This substitution of mechanical/system complexity with intelligent algorithms achieves high prediction accuracy while maintaining manageable system complexity.
Data Source
AI summary
An elevator system is provided. The system comprises a first acquisition unit that acquires an image in a car of an elevator and an image of a landing of the elevator; a first generation unit that generates first learning data from the acquired images; a learning unit that performs learning using the first learning data, thereby generating a first learned model; a second generation unit that generates input data from a new image; an estimation unit that estimates, by applying the input data to the first learned model, whether the person on the landing of the elevator, which is included in the new image, gets in the elevator; and a control unit that controls an operation of the elevator based on an estimation result.