We’ve all enjoyed the thrill of getting on the elevator for a quick ride from the hotel lobby to our room on the 40th floor… only to realize that people are getting off on the 5th, 8th, 12th,… well you get the picture. By the time you finally reach your floor angry looks are being exchanged with the other passengers, almost daring them to push a button to stop the elevator. In the morning the reverse happens… the elevator makes 40 stops on the way back down to the lobby, filling up by about floor 20, and people on lower floors end up catching the up elevator in order to go down.
The Marriott Marquis on Times Square in New York City apparently had a similar, albeit even more severe, problem. I’m not sure if this also sparked the recent addition of artistic metal grates above the interior balconies; perhaps someone with a Starbucks addiction gave up waiting for an elevator and tried a quicker route down to the lobby. However the elevators at the Marquis are rather unique.
Instead of simply getting on the nearest open elevator with everyone else, you type your desired floor into a keypad. The system then assigns you to one of fourteen elevators. You go and wait a few seconds, the elevator arrives, and you get on along with anyone else with the same assignment. The first thing you notice is that there are no floor buttons inside the elevator, with the second observation being that a small LED readout shows your floor along with at most one other. Up you go with no more than one stop.
Presumably the system assigns you to an elevator along with other people with a similar or at most one other destination, taking into account number of requests and the time until the next elevator arrives. Perhaps there’s even some portion of the algorithm that aims to equalize wear and tear on the various elevators. But the result is a quick ride to your destination.
Dare I say that software helps expedite this process, and perhaps this is a non-manufacturing application of “part family” algorithms? There’s probably a visual controls method that could achieve some of this efficiency, but not all. After the tizzy I created on the JSLEAN discussion group earlier this week with my itsy bitsy sideways comment on them, I wonder how they’ll react to this admission.
mike says
I think its interesting how some transportation systems are pull oriented like elevators and some are more of a scheduled push type.
Now I believe hypothetically for the sake of argument you could create elevators and a visual controls method that would completely optimize the elevator paths in a system of that complexity, but it would be impractical to implement requiring highly trained elevator operators and or passengers using lots of real time visual data and meomorized standard visual solutions and or lots of mental calculation.
I think visual control ‘methods’ are algorithms too. However they are simple and flexible by ones by necessity of implementation. I think they are often cheaper and more reliable to use and quicker to implement also. They often attach data and info to physical objects people and machines in the physical workplace so important information doesnt get missed out on during design or operation by either the control sytemn or the people working in it with it part of it, or for it. I also like the fact that workers with the most experience can easily be part of the visual controls method design and the continual upgrading of it.
david foster says
An interesting post on the user interface to elevators, here:
http://blogs.business2.com/design/2006/10/how_to_make_a_s.html
Also, I did a post on the floor pre-selection elevators, and the sociology thereof.
http://photoncourier.blogspot.com/search?q=elevator
Karthik Chandramouli says
I’m interested to see reactions from anyone who has actually stayed at this hotel and used these elevators.
I have.
And I hated it.
And I’m not the only one.
As much as it appeals to the technology geek in me, IMHO, it’s a classic case of adding value for which the customer will not pay.
In the Toyota Production System, we try to optimize processes around the people, whereas conventional thinking drives us to optimize processes around expensive machines.
To me, this “smart elevator” glorifies the machine at the expense of the person. We implement sophisticated queueing algorithms to optimize the utilization of elevators (expensive capital assets).
The noble goal of reducing wait times for hotel guests to reach their destinations comes in as a distant second priority.
How effective is this system against its primary goal (asset utilization) vs. its secondary goal (shortest lead time for the customer)? I would love to see the data from Otis Elevator.
And then, please show me the customer satisfaction data that goes along with it…
From the perspective of the hotel guest (the customer), it’s difficult to see the value-added in this process. Let me illustrate by example:
This hotel is an extremely popular location, situated in the heart of Times Square in New York City. As such, it’s always full of conventioneers and other meeting or event attendees.
When the hotel gets busy, invariably elevator usage creates congestion and excessive wait times.
No doubt, this is why the owners of this hotel decided to fix the problem.
Here’s what happened to me.
We went for a night cruise as part of our event, and came back to the Marriott Marquis late in the evening.
In the lobby, we punched in our destination floor and waited for the elevator to arrive. At least 50 other people came in on our heels and punched their destinations into the keypad.
There is no clear visual indicator that my request had been received and processed, or that it had been prioritized vs. other people who had arrived after me. There is no indicator of when the next elevator is arriving.
As an aside, should we transport passengers FIFO (or in this case, first-come, first-serve), or batch them to optimize the elevator’s capacity utilization?
The human interface on this “smart elevator” is not intuitive, so each person waits a short time, and then re-enters their floor request, just to make sure the request is processed.
This is the equivalent of pressing the already-lit “up” or “down” button when waiting for an elevator to arrive. But the effect is far more deleterious.
In the latter system, pressing an already-lit button has no effect, except to assuage the passenger that the elevator is, indeed, on its way.
In the former system, re-entering a floor adds demand for another destination to an increasingly congested queue. (Can you game the system by entering your floor many times?)
Smarter people than I can explain the negative effects of congestion in a queue, but needless to say, the elevator lacks intelligence to filter out this “phantom demand” vs. a genuine request by another passenger.
Not only is the interface poorly designed, but these systems are not widely deployed, so users tend not to have much confidence in how the system is supposed to work.
So the hotel staffed the area with people who explained how it worked. To me, this is total waste because they could not do much more than reassure the guests that the elevators will eventually show up.
Finally, the elevator came, and we were on our way up to the room. It was then that I realized that we needed to stop on an intermediate floor and pick up our suitcase from a friend’s room.
Unfortunately, there’s no way to stop the elevator once you’re on your way. There’s no buttons inside to change destinations en route.
Not only is there no “andon” when you’ve made a mistake (or, in this case, changed your mind), it is completely unnerving to be at the mercy of the algorithm.
In short, this is an example of a good idea with poor execution. The road to hell…
More than a year ago, this issue was covered by the New York Times:
An Elevator to Your Floor, With No Local Stops
http://www.nytimes.com/2005/11/02/realestate/02elevator.html?ex=1182312000&en=4b53086df2be7855&ei=5070
I’m sure they have improved the experience from when I first used it (just after it was installed). But I wonder if they are just expecting people to modify their behavior and adjust to the machine?
Kevin says
Actually I was there all last week, which prompted the post. Your comments are well-taken.
I also noticed many confused people who had not experienced that system before… in fact it was rather amusing seeing people get on the elevator and then panic when they couldn’t press a button for their floor. You’re also right in that there is no way to “divert” the elevator once on (perhaps a good thing for the other passengers?). However I did notice that when a large number of people were going to the same floor, and I happened to be on the very popular concierge level on 30, a second and then third elevator were called as more people punched in that floor. However often only one person in a group would punch in a floor, so the elevator wasn’t smart enough to figure out how many people were in the group, leading to over- and under-laden elevators.
But the comments I heard from people that stayed at the Marquis before the new system was put in place made me feel that it was a huge improvement. I heard several horror stories about waiting over 30 minutes, walking up/down 30 flights to beat the elevators, etc. Poor initial design perhaps.
Still some flaws, but definitely an improvement.
Kevin
Mark Graban says
I’ve stayed here and echo Karthik’s points on the confusion, lack of good user interface, the poor visual controls, and the waste of needing to staff people there to “help.”
That said, I also heard anecdotal data from guests who said “the waits used to be MUCH worse” with the old traditional elevator system.