Traditional elevator systems operate on predefined logic, responding passively to button presses without considering real-time usage patterns or user preferences. This outdated approach leads to inefficiencies such as longer wait times, poor resource allocation, and a lack of personalized service. Our client, ZREMB, sought to revolutionize elevator design by incorporating AI-driven decision-making to create a more dynamic and personalized experience for residents.
The key challenge was developing an intelligent software system that could analyze elevator usage patterns, optimize operations, and enhance user interaction while working within the limitations of real-world installations. The solution needed to function efficiently even in buildings with weak internet connectivity and limited computing power. Additionally, ensuring privacy was crucial, requiring data anonymization techniques to process user interactions without compromising personal information.

To address these challenges, we developed a machine learning-powered elevator system capable of predictive decision-making, pattern recognition, and adaptive functionality. The system was designed to learn from usage data, detect behavioral patterns, and optimize movement to minimize wait times and improve efficiency.
Computer vision technology was integrated into the system to enhance user interaction and personalization. The camera-based system could recognize patterns in people movement and usage, allowing for automated elevator control while maintaining privacy through data anonymization. To ensure compliance with data protection regulations, all collected data was processed in a way that removed any personally identifiable information.
Given that many facilities had limited bandwidth and computing power, we optimized the system’s algorithms for efficient processing. This included reducing the computational load on local hardware and ensuring that data transfers were minimal, enabling smooth operation even in resource-constrained environments.
Before deployment, rigorous testing was conducted, including real-world validation in an operational facility. This allowed us to refine the software, ensuring reliability and responsiveness under real-life conditions. The final implementation provided a seamless, intelligent elevator experience, aligning with the client’s goals for modernization and innovation.

The project was delivered successfully within the planned timeline and budget, first with a functional MVP that met all performance requirements. The intelligent elevator system significantly enhanced efficiency and user experience by reducing wait times and optimizing travel paths based on real-time data. The implementation of computer vision provided a more adaptive, personalized service while maintaining strict privacy compliance.
As a testament to its innovation, the Martha Elevator, developed as part of this project, was recognized at the PropTech Festival 2022, ranking among the top three technology solutions in real estate. The client’s appreciation of our clear communication and structured development approach was reflected in smooth collaboration, with regular progress meetings ensuring alignment at every stage. During implementation, we provided ongoing support, quickly addressing any necessary adjustments to refine the system further.
By integrating AI and machine learning into elevator design, this project demonstrated how technology can transform even the most traditional systems. The result is an intelligent, user-centric elevator experience that enhances convenience, efficiency, and security while maintaining compliance with data protection standards.

We began by analyzing the client’s vision and transforming it into a detailed product specification. The next step involved selecting and configuring the appropriate hardware, ensuring compatibility with the software’s requirements. Our team designed a scalable management system capable of processing real-time data efficiently while maintaining security and compliance.
To enable advanced functionality, we developed custom computer vision algorithms that allowed the system to recognize movement patterns and optimize elevator operations. The algorithms were carefully optimized to run within the constraints of limited computing power, ensuring smooth performance even in environments with weak internet connectivity. Real-world software verification was conducted in a live elevator setting to validate performance before final deployment.

The system incorporated a combination of classical machine learning models and deep learning techniques for predictive analytics. Advanced computer vision methods, including motion detection and object tracking, were used to analyze movement patterns. Data anonymization techniques were implemented to ensure privacy compliance, removing any personally identifiable information before processing. Optimized software architecture allowed efficient execution on limited hardware, minimizing processing delays and bandwidth usage.