Top Page > Introduction to On Demand Buses
Introduction to On Demand Buses

1. What is an On Demand Bus?

On Demand Buses are buses where people ride-share using a reservation system. Unlike buses that travel a fixed route on a fixed schedule, on-line buses have no fixed route or schedule. Just like with a taxi, the bus goes just where the passenger wants. What's different from a taxi is that passengers ride together. Passengers traveling similar routes ride together, with the bus making only short detours along the way. In the case of Tokyo University On Demand Buses, even in cases where there were slight detours for people sharing the ride, it was possible to stay on time, arriving at the time promised to the passengers booking the ride.

On Demand Bus
Route Bus
Not Needed
Not needed
Time Schedule
Designate the time that works for you.
Run according to the time table.
Designate the time that works for you.
Getting On and Off
The bus travels the shortest distance between stops.
Buses stop at every stops along a predetermined route.
Taxis travel the shortest distance between two desired points.
In-between that of a bus and a taxi

2. Existing On Demand Buses

On Demand Buses are an extremely efficient form of transportation. That's because, unlike with fixed route buses, On Demand Buses don't have to waste time going to stops where there are no passengers. On Demand Buses take the fastest route where there actually are passengers. Because of the efficiency of these kinds of On Demand Buses, they have come to be used the past five years in Europe, the United States, and Japan. In the earlier period in Japan, they were operated in Kouchi Prefecture and Fukushima Prefecture, and they have come to be used in more and more places, especially in local areas.
However, while On Demand Buses have been in operation for some time, there have been few examples of the system being used in a region where it would be used by many people. For example, normal route buses (community buses) are mainly used for morning and evening commutes by passengers commuting to work and students commuting to school. However, On Demand Buses have tended to be used only by housewives going shopping and elderly traveling to hospitals, with few tests of their applicability to school and work commutes.

3. Existing problems: the use of operators and delayed arrival times

Our project started out in the hopes of determining why On Demand Buses public transportation systems have difficulty infiltrating new regions. We came to the conclusion that there were two main factors: 1) the fact that users must call an operator, and 2) the fact that buses arrived late. These are detailed below.

The first problem is that passengers must call an operator. With current On Demand Buses, the passenger calls an operator and places a reservation. This means that ach time a reservation is made, a scenario like the one below takes place.

Hello, my name is Suzuki. I would like to take a bus from my house to the hospital at 10 o'clock.
I understand, just a moment, please.
Fine, we will have a bus at your home at 10 o'clock. Please wait there for the bus.
Thank you.

While there are users who cherish the fact that they can place a reservation by phone, there are no doubt many others who find having to place a call for each reservation tedious. Moreover, if people have to place a call to an operator to make a reservation, then in towns with large populations, it would be necessary to set up an extensive call center, leading to higher operating costs.

Another problem is that of late arrivals. On Demand Buses take on reservations one after another, and if its route should change along the way, it may be rather difficult for the bus to arrive on-time. Take, for example, a case where?at the time the first passenger made a reservation?the bus is set to pick her up at 10 o'clock so that she can reach her destination at 10:30. In this case, the bus is booked to arrive at the person's home at 10 because it takes 30 minutes to arrive at the hospital, where the passenger needs to be at 10:30. Now, what if another passenger makes a reservation, saying "I need to arrive at the hospital at 10:30."
Before the new passenger places a reservation, the route is a direct one from origin to destination, like the dotted line shown in the figure. But after the reservation from the second person comes in, then the bus must make a detour so that the two passengers can share the ride. This is precisely what causes the delay. And the more passengers there are placing reservations, the longer the delay becomes. Consequently, any vehicle where you don't know how long it will be delayed will not be used by people who don't have time to spare.

4. Characteristics of the Tokyo University On Demand Bus

Tokyo University On Demand Buses resolved the current issues with On Demand Buses described above by using computers. Tokyo University On Demand Buses have two distinguishing characteristics: 1) there is no need for operators, and 2) it incorporates a procedure for creating schedules where the arrival times will be met.
First, the computer automatically handles the reservation, which can be sent over the Web, by email, or with an automatic voice operator. Second, the route is automatically determined after considering various parameters. And third, the computer automatically handles the operation of buses that are already running. In addition, using the latest calculations by the computer, the system determines route plans automatically in a way that safeguards arrival times.
The calculation algorithm developed by the Tokyo University research team performs calculations on the spot, looking to add new passenger reservations only to the extent that they will not interfere with reservations already established for passengers with previous bookings. Thanks to this special calculation method (algorithm), it has been possible to realize an On Demand Bus system that is processed fully automatically without delays.

5. Operator Zero Project

It was pointed out that one of the characteristics of the Tokyo University On Demand Bus is that operators are not needed. But technically speaking, that is untrue. In order for there to be no need for operators to field the Tokyo On Demand Bus reservation system, all users must use the Web, mobile phones, email, or push buttons to place a reservation.
With existing On Demand Bus systems, one of the essential requirements was that users place a call to an operator. With the Tokyo University On Demand Bus system, so long as you are capable of using the bare minimum functions of a personal computer, mobile phone, or email service, you can easily make a reservation. But for those who are unused to using IT devices and who struggle to perform even the most basic tasks on a personal computer, mobile phone, or email service, then placing a reservation is decidedly difficult.
For that reason, even in the trial system operated by Tokyo University in Kashiwa, we set up a simple call center and hired a single part-time operator to handle calls. Existing On Demand Buses have two flaws: 1) the more reservations there are, the more operators you need; and 2) operators need to have special training so that they can dispatch buses taking complex routes into consideration. By contrast, in the case of the Tokyo University system, since you or someone in your household with a personal computer can place a reservation, with respect to the first concern (1 above) you don't necessarily need more operators as the number of reservations increases. Moreover, since the operator is simply entering information onto a web page that is right in front of her, the computer handles all route creations and bus dispatch instructions. Thus, with respect to the second concern (2 above), there is no need for the operator to have any special skills or training in this regard. In past experiments conducted by Tokyo University, we handled reservations for normal business matters, including nurses having fixed appointments at hospitals or city hall employees having meetings at institutions.B

Given this opportunity, one of the projects currently being undertaken by the Tokyo University On Demand Bus research team is the "Operator Zero Project." The goal of this project is to realize a system where operators are unnecessary. In other words, we want to make people who find it difficult to place reservations under the current system capable of making reservations by themselves. We haven't yet realized "operator zero," but we would like to discuss some of our results so

1) Development of a simple reservation terminal
By narrowing down the functions to obtain a reservation, the system is such that a reservation can be made as easily as using an automatic ticket machine at a station or an ATM terminal at a bank. This was introduced as one of the programs on regional broadcasts in digital television, so that reservations could be made using a TV remote control. Alternatively, a terminal could be placed at hospitals, stations, or super markets so that users could place reservations for On Demand Buses at practically any street corner. (The figure on the right is a screen shot image of the terminal system.

2) Development of booking suggestion email system
The booking suggestion email is another of our efforts for the "Operator Zero Project." We analyzed previous use conditions and were able to make predictions at what day and time passengers would want to travel. Putting this into an email, we would send it out to families or friends who could use email. The person receiving the email could then click on a button and set up a reservation simply by replying to the email.

Operator Zero has not been fully implemented yet, but we are conducting proof of concept experiments in each area so that it can be realized in the near future. It should be available very soon.

6. How cheaply can Tokyo University On Demand Buses operate?

The Tokyo University On Demand Bus uses a shared server format (ASP service), meaning that the system can be operated inexpensively as there are no additional expenses incurred for introducing separate systems, upgrading the system, or conducting system maintenance.
When existing systems have been actually introduced, for the most part they were purchased as a whole system by the government (figure at left). By contrast, with the Tokyo University On Demand Bus system, we adopted a "service purchase format", and so compared to existing systems, we are able to greatly reduce the costs typically associated with upgrades, operation and maintenance, and the introduction of new components. In addition, not only is it possible to provide safe services from a security angle or server management by special organizations, it is also possible to always obtain the latest in services. To see how much it costs to operate the system, click (here).
However, the reason the Tokyo University On Demand Bus system is so cheap is not so much because we want local governments to operate the system without incurring much cost, but rather because we have striven to introduce the most suitable format.
When we decided to introduce a new form of public transportation, we felt it was critical to focus heavily on operating costs. It would be reckless to introduce a system that could not operate long-term or that had no conditions on its income and expenditure model. It is important to work out a mode of transportation that is sustainable.
However, it wasn't sufficient to focus simply on costs, as performing accurate pre-tests and demand predictions, and hammering out a robust introduction plan were equally important. Speaking to those in local governments, we were often told the following:

Thanks to the introduction of On Demand Buses, compared to community buses that were for all intents and purposes "transporting air," subsidies were about half of their former levels. Even so, officials were still struggling to meet the subsidies and had no funds to spare.
The charge for community buses was a uniform ¥150 per ride. Three years earlier they had performed a survey of residents and determined that an overwhelming number of respondents said that even at ¥200 per ride they would prefer to use a bus over driving a car. But even though officials were charging ¥50 yen less than that, still, at ¥150, the buses were basically empty.
Even though the population increased only 3% when consolidating cities, towns, and villages, town planning regulatory boundary areas were tenfold or more. Although we have provided a community bus service that serves as a transit up to the main trunk line buses, ridership has only scarcely increased.

There is no question that there are many situations where the Tokyo University On Demand Buses could be operated at far less subsidized cost then empty community buses, and that from a cost perspective they would be far more efficient. However, we would also need to set aside a budget so that when they were first being introduced, we could listen carefully to residents, work out an introduction plan that was appropriate in the long term, and following introduction, obtain feedback on whether they were being used properly.

7. Costs associated with operating On Demand Buses

7-1 Fares charged for the On Demand Bus reservation and bus dispatch system

The costs associated with using the Tokyo University On Demand Bus system break down into 1) server administration costs and 2) rental fees for on-board devices. There are no charges whatsoever for initial system introduction or system upgrades.

(1) Monthly server administration costs

The monthly server administration costs were calculated according to the method below. Essentially, the cost changes according to the amount of demand per month (namely, the number of users). Please see the formula below.

Take for example a case where there were 1600 passengers in a month. Then the actual demand value is 1600 people.

Since actual demand value is 1600 passengers, which is between 1000 and 9999 passengers, then (looking at the chart) we see that the demand ranking is 2. That means that the basic management cost will be ¥60,000, which shows that the free-of-charge demand figure will include the 1000 passenger demand portion. A demand of 1600 passengers is only 600 riders over the free-of charge demand value, so to calculate the basic management cost you add 600 * 5 = 3000.

Taking actual demand value to be 1600 passengers and applying it to the above formula, we obtain the following:

Server administration costs = ¥60,000 + (1600 - 1000) * ¥5 = ¥63,000

1 month Total number of passengers
Basic management cost (monthly)
Portion in excess of free-of-charge demand (per rider)
999 passengers
1,000 free-of-charge demand
9,999 passengers
1,000 free-of-charge demand
99,999 passengers
10,000 free-of-charge demand
999,999 passengers
100,000 free-of-charge demand
9,999,999 passengers
1,000,000 free-of-charge demand
10,000,000 passengers
10,000,000 free-of-charge demand
Free-of-charge demand value
Basic management cost
Demand units

(2) Rental fees for on-board units

Now we will explain the charges for on-board units. First, the rental charge for on-demand units will depend on the vehicle capacity (excluding the driver). In addition, it also depends on whether the units are being used for a proof-of-concept experiment or actual operation.

If the vehicle capacity is less than five passengers (excluding the driver), in other words, if the vehicle is a sedan or something similar, then costs are low. This is because the on-board unit's specifications are different. For capacities of 9 passengers or more, then the on-board unit must be the size of a personal computer, whereas for capacities of 8 passengers or less, the unit can be the size of a PDA.

The on-board unit charge breaks down into a monthly communication charge and a monthly maintenance fee. First-time installation fees are needed based on the description in the installation manual, but someone in charge can be dispatched for that. There will naturally be a separate fee charged for installation.

Rental Fee for On-Board Units (When Used in Proof-of-Concept Experiments)
Passenger Capacity
Experiment Period
Rental fee per unit *
9 or more
Less than one year
¥60,000 per month
Less than two years
¥40,000 per month
Two years or more
¥25,000 per month
8 or less
Less than one year
¥16,000 per month
One year or more
¥13,000 per month
*The rental fee includes communication costs and maintenance fees.

The rental fee is independent of the actual time of operation. It does change depending on vehicle size as shown below. The fees in the table below include the communication fee and the maintenance fees.

On-board unit rental fees (When used for actual operation)
Passenger Capacity
Hours of operation
Rental fee per unit *
9 or more
The length of operation has no effect on fees.
¥25,000 per month
8 or less
The length of operation has no effect on fees.
¥12,000 per month
*The rental fee includes communication costs and maintenance fees.

Because of the shared server format there are no initial costs and there is no annual system upgrade costs. Communications expenses and maintenance fees are also included. In other words, On Demand Bus operation can be achieved with only the above system costs.

7-2: Supplemental Items (Operator, simple reservation terminal) *Optional

As said earlier, we can imagine that there are passengers who will find it difficult to make a reservation on their own using IT devices like mobile phones, personal computers, fax machines, and email. For such people, a simple call center should be established. In addition, it is possible to establish simple reservation terminals at supermarkets, health centers, community centers, and hospitals. These simple reservation terminals would amount to "street corner" terminals that are activated by touch screens and that have built-in voice response.

[Operator fee]

* With respect to operators, we suggest that local government and administrative bodies each establish simple call centers.

* Just as with normal reservations, operators need to be able to listen to users' needs and enter that information into web pages. They do not need special technical skills.

[Simple reservation terminals]

* Simple reservation terminals provide their service through Internet lines. While there are recommended devices for this purpose, these terminals can be set up anywhere so long as we provide 1) touch screen monitors, 2) printers, and 3) Internet-capable personal computers for each.

* To use the terminals Internet lines are needed (1 Mbps or greater speed are recommended).

8. Optimal introduction plan

Introducing On Demand Buses at Tokyo University required us to perform the following five steps:

1) Income and expenditure balance simulation
2) Simulation of the effect of introducing the system
3) Proof-of-concept experiments
4) Evaluation and design modifications
5) Actual operation

8-1: Income and expenditure balance simulation

First, income and expenditure balance is estimated. Click here to download the Excel sheet. (Japanese and English only.) The upper cells on the Excel sheet in red indicate 1) how many vehicles can be procured, 2) whether operators will be set up, and 3) what the management costs are.
The yellow cells at the upper portion of the Excel sheet indicate 1) how level of subsidies needs to be spent, and 2) what the estimated advertising expenses will be. The figures in the table below change automatically. This enabled us to see how many vehicles would need to be operated and how many fares would need to be received, taking the need to operate the system on a sustainable basis.

8-2: Simulated results of system introduction

Using a complex computer simulation, we ask ourselves what kind of service could be realized when operating the system using the number of vehicles sought in 7-1. This simulation is performed for free by the Tokyo University research team. In order to create the input data for the simulation, it is necessary to investigate the regional traffic conditions, getting people to respond to questionnaires like this one.

8-3: Proof-of-concept Experiment (1 - 5 months performed over two separate periods)

If both the income and expenditure simulation and the operation simulation meet the preconditions, then we know that the system is sustainable. At that point, we try operating the system in earnest. The point of this step is to see whether the operation goes as expected based on the simulation, as well as whether there are any region-specific issues? In most cases, proof of concept trial runs have been conducted without fares, owing to the fact that they were subsidized by the prefectural or central government. The duration of the trial depends on the budget, but to be meaningful, trials must last several months and be conducted over two separate periods.

8-4: Evaluation and design modifications

By analyzing the trial run midway, it is possible to determine whether the On Demand Bus system is truly effective for the region in question. We will now give an example of the evaluation standards that can be applied at that time. Moreover, we will also give an example of a case where the simulation requirements were modified in light of trial run results, together with areas to be improved for the successive trial run.

  • How effective was the system compared to existing policies like free taxi tickets or subsidized community buses judging from factors like rate of operation, ride-share effectiveness, service level, and economic result.
  • Establishment of zones that avoid competition with existing transportation systems
  • Sustainability of service and subsidies (evaluation of uncertainties in various areas)

8-5: Actual operation

The results of the proof of concept experiment agreed with the usage conditions hypothesized by the simulation. If one knows that the system is sustainable, the next stage is to implement a real operational system. However, periodic evaluations need to be undertaken.

We have described in simple terms the steps needed to take up to system introduction. We feel that the paramount issue is that of being able to establish a sustainable transportation facility. Sustainability is something that must be judged while taking into consideration various uncertainties like expenses, changes in rider usage conditions, fluctuations in regional population, and the like.

9. Decisions to be made when operating On Demand Buses

Summarizing what we said above, determining whether On Demand Buses can be successfully introduced amounts to nothing more than resolving the questions below. We looked into this issue while also seeking to determine the most appropriate form of on-demand service under the circumstances.

  • What is the target number of passengers per day? Is that figure based on the right foundation?
  • In order to transport the target number of riders, what mode of operation would be suitable? What form of service should we design?
  • What operational system should we have?
  • Are telephone operators needed? If so, whom do we hire?
  • How do we procure vehicles?
  • At what level do we set passenger fares?
  • If subsidies are needed, how much assistance is needed?

  • Lastly, is the above-described operational system sustainable? If conditions change slightly, would the system no longer be sustainable?

Many parties have expressed an interest in the On Demand Bus system set up by our laboratory, including governments, taxi companies, bus companies, and construction companies. For those who are interested in our system or who might like to introduce it and would like to know more, please contact us using the contact means described below.

10. Contact information

Tokyo University Graduate School of Frontier Sciences, Human Environmental Studies Major, Industrial Environment Studies Area, On Demand Bus Research Team

Address: Environment Building Rm. 283,
5-1-5 Kashiwanoha, Kashiwa-shi, Chiba-ken 277-8563, JAPAN
Tel. 04-7136-4629
Fax. 04-7136-4628