AI Vision Door Control for Accurate Approach Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Building Management Systems (BMS) for automatic doors have limited operational ranges and rely heavily on single cameras or motion detectors, providing a limited secure door automation experience in smart environments.
Innovation Solution
An AI-based system that processes vision data using separate computer vision and decision-making models to accurately determine when to activate automated doors, considering factors like presence, speed, trajectory, distance, and face orientation, with adaptive threshold distances and real-time learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single camera or motion detector is used in existing BMS systems, then the system structure is simple, but the operational range is limited and detection accuracy is insufficient
Solution Approach 1:
The patent combines multiple vision data capture entities (cameras, depth sensors, motion detectors) into a unified door automation system. These diverse sensing devices work together to capture comprehensive vision data about approaching subjects, enabling accurate detection of presence, distance, speed, and trajectory while maintaining system manageability through integrated processing.
Solution Approach 2:
The vision data capture entities are designed to perform multiple functions simultaneously - detecting presence, measuring distance, tracking motion trajectories, and determining subject orientation. This multi-functionality allows the system to achieve high detection accuracy across multiple parameters using a unified sensing architecture rather than separate specialized devices.
2Speed
If the door controller is located within a short range from the automated door actuation device, then the response time is fast, but the operational range is limited
Solution Approach 1:
The patent transitions from relying solely on physical proximity (spatial dimension) to detecting subject approach through multi-dimensional vision data analysis. By capturing and analyzing trajectory, speed, distance, and orientation data from multiple capture entities positioned at various distances from the door, the system can trigger door activation based on predicted arrival rather than immediate proximity, thereby extending operational range while maintaining responsive activation.
3Reliability
If traditional BMS systems are used, then the system is easy to operate, but the door automation experience is limited and security is insufficient
Solution Approach 1:
The system performs automated analysis of vision data to determine subject intent and predict door activation needs without requiring user configuration or intervention. The machine learning model automatically processes multi-source vision data, extracts relevant features, and makes intelligent decisions about door activation timing and parameters, providing secure automated operation that adapts to different users and scenarios.
Solution Approach 2:
The system continuously monitors vision data from multiple capture entities and uses machine learning models to analyze subject behavior patterns. This feedback loop enables the system to learn from observed trajectories, speeds, and orientations, improving its ability to predict when doors should be activated and enhancing security by distinguishing between genuine approach intentions and passing motion.
Data Source
AI summary
A system for an automated processing of vision-based data for an automated door controller, including a processor of an automated door control server node configured to host a machine learning (ML) module and connected to at least one vision data capture entity and to at least one target controller of the automated door over a wireless network connection and a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to: acquire vision data from the at least one vision data capture entity reflecting a subject in a vicinity of the automated door, the vision data comprising a signal; normalize the signal for depth consistency; extract a set of classifying features reflecting movements of the subject relative to the automated door from the normalized signal; generate a classifier based on the set of classifying features; provide the classifier to the ML module configured to generate a predictive model based on a neural network for producing at least one decision parameter associated with an operation of the automated door; generate a command for operation of the automated door based on the at least one decision parameter; and send the command to the target controller.