| Summary: | 碩士 === 國立臺灣海洋大學 === 電機工程學系 === 105 === Due to industrialization of our country, the number of automobiles, especially passenger cars, has continued to grow for many years. The rapid increase of motor vehicles has created many traffic related problems for our society; parking difficulty is one of them, especially in urban areas. To alleviate the traffic and parking problems, government agencies often urge the general public to stick with public transportation systems as much as possible. While public transportation systems are generally cheaper to use for daily commute, they offer much less or no flexibility in both schedule and route. Furthermore, when commuters need to switch among different transportation systems such as the bus and the municipal rapid transit, the additional time costs stack up. In recent years, a number of cities in the world chose to develop or experiment with a new type of bus system, often called on-demand bus. While differences exist, on-demand buses tend to have flexible route and/or schedule. On-demand buses attempt to combine the low-cost characteristic of the traditional bus and high flexibility of the taxi. Under right combination of population, city size, and quantity of service vehicles, on-demand buses can offer convenience far surpassing the traditional bus can provide, especially for the following categories of passengers: senior citizens, people with physical disabilities, outpatients traveling between hospitals and homes, tourists unfamiliar with the city.
In this thesis, we have developed an experimental on-demand bus system based on Google Cloud Platform (GCP). The cloud-based service is developed with Google App Engine (GAE). The processing begins with translating the source and destination addresses of service requests to their corresponding latitudes and longitudes through the Google Maps Geocoding API. All pending requests will be analyzed to determine whether or not a request should be included in a route. Then, the accepted requests are sequenced to generate a shortest route satisfying the current combination of requests. Upon the completion of path planning, the route and expected pick-up and arrival times are displayed on the service Web page to inform the customers. Current implementation allows only one service vehicle running, but the algorithm for multiple service vehicles has already been devised and discussed in the thesis. Because of the extensive use of versatile Google APIs, our current system is able to avoid congested streets and can pick up or drop the passengers on the correct side of the road.
To verify the feasibility of the on-demand bus prototype, we use 500 sets of service requests for on-demand bus, 100 sets for car (simulating taxi) and 100 sets for traditional bus to perform the simulation. The experimental results show that the operation of on-demand bus largely conforms to the underlying notion that both the time to fulfill a request and the cost of ride are between those associated with the traditional bus and the taxi.
Keyword & phrases: On-demand bus, cloud computing platform, Google APP engine.
|
|---|
