Introduction
Elevators are ubiquitous in our daily lives, whether in office buildings or residential towers. Yet, despite their apparent simplicity, they hide a fascinating algorithmic complexity. Who hasn't cursed while waiting for an elevator that never seems to arrive? Let's delve into the mechanisms behind these metal boxes.
Early Algorithms: SCAN and LOOK
The SCAN algorithm, patented in 1961, is one of the most primitive systems. The elevator starts at the ground floor, goes up to the top floor, and then descends, stopping at the requested floors. Simple and effective for low-rise buildings, it quickly shows its limitations in larger structures.
Then comes the LOOK algorithm, a more efficient variant. Instead of always going to the top floor, the elevator only goes to the highest requested floor, saving time and energy.
Complex Coordination: Multi-Elevator Systems
In buildings with multiple elevators, coordination becomes key. The most basic method uses a central scheduler that assigns requests to the nearest elevator. However, this approach can be optimized.
Advanced Algorithms: The RSR Algorithm Example
Otis' RSR (Relative System Response) system innovates by scoring each elevator based on its suitability for each request. This helps optimize not just proximity but also capacity and load.
Measuring Efficiency: Wait Times and Statistics
How do you measure the efficiency of an elevator algorithm? Wait time is a key indicator. For example, a p90 of 2 minutes means 90% of users wait 2 minutes or less. The goal is to reduce these wait times to improve user experience.
The Challenges of Peak Periods
Passenger flows vary throughout the day. In the morning, elevators are swamped by those going up, while in the evening, it's the opposite. Lunch periods see flows in both directions. Adapting algorithms to these variations is crucial to avoid excessive wait times.
Future Innovations: Towards Smart Elevators
Smart elevators use sensors and artificial intelligence to anticipate user needs. For example, by analyzing usage patterns, an elevator can predict peak periods and adjust its algorithms accordingly.
Conclusion
Elevators are more than just vertical transport means. They are at the crossroads of mechanics, computer science, and artificial intelligence. Every innovation aims to improve our comfort and efficiency in everyday life.
Let's discuss your project in 15 minutes.