top of page

Senior Design Project

Intelligent System for Optimizing School Bus Routes

Based-on Shortest Travel Time

Department of Electrical and Computer Engineering, King Abdulaziz University, Saudi Arabia

RAGHAD AL-SOLAMI, RAWAN AL-JABALI, RUAA OBEID, RUBA BIN JABAL

 

ABSTRACT

Amid Saudi Arabia’s 2030 vision, over the past three years, the Ministry of Education has planned for school transportation to cover more geographical areas and provide transportation services that are safe, cheap, and convenient. Hence, route optimization is a crucial factor in enhancing and improving transportation services. With today's technology, route optimization can be efficiently done using Artificial Intelligence (AI) algorithms. It helps in improving delivery time, reducing fuel costs, and increasing customer satisfaction. Manual and inefficient scheduling of school bus routes locally causes high costs and lengthy traveling time of vehicles. In addition, poor emergency management when school buses are en route. The objectives of this project are to develop an AI-based route optimization algorithm for school transportation that reduces the travel time delay. Besides, provide dynamic route optimization that adapts to emergency conditions relating to the driver, students, and bus. In order to achieve those objectives, the Simulated Annealing (SA) algorithm will be utilized with the assists of Google Maps APIs and Firebase Google Cloud to deliver a mobile application for school bus providers. The output will be displayed through an LCD screen that is connected to a Raspberry Pi board, which will run the route optimization algorithm. In turn, the board will be connected to a GPS module that will be tracked by Google Maps API, backed up by the Firebase cloud database.



SYSTEM DESCRITION


The chosen baseline product is a multi-destination route optimization system. The system aims to serve schools and school bus transportation providers with an efficient route planner for their students. The optimized route will be given with the shortest traveling time for the overall bus route. Schools bus transportations providers will enter the following: students' information, which includes name, ID and their house location, number of school buses, the capacity of the bus, and the school's location. Then, students will be classified based on their locations and assign them to the buses while taking into consideration the bus capacity. Following that, the system will compute the optimized bus route with the shortest traveling time and project the route into the map on a mobile application. The bus driver will then follow the shown route on the map through the user interface. A second aim of the system is to increase student's safety by providing an emergency option. In case of an emergency, the bus driver will click on an emergency button, and the system will immediately add a new location to the current route based on the emergency case. If the case was to go to a hospital, for example, then the system will identify the nearest hospital to the current bus location and add it to the route. A predefined set of action upon emergency include either adding the nearest police station, hospital, or car repair.



RESULT EXAMPLE

The system was experimented with a set of data, and the input parameters were determined as follows.

  • Number of Locations = 19

  • Number of Bus = 2

  • Bus Capacity = 10

  • Demand = 1

First, the user interface was used to enter the data as specified before. As shown in below Figure, each student information was passed to the database through a form in the website by the user.

From the Firebase side, the two Figures below illustrate the data was successfully uploaded in real-time to the database.

After that, the optimization program was executed with those data. It generated a set of routes for each bus, including the number of students and time interval of each route.The optimum routes were represented with Google Maps, demonstrating the visiting order of each route, as shown in below Figure. It is apparent from the generated result that the optimization program was distributed to the students among the buses efficiently according to their locations. The solver status returns the value of one as the status of the search, which means the problem was solved successfully according to OR-Tools.


The total number of generated solutions from the search log was 1469, where the optimization algorithm traversed through it with a runtime of 29,995 ms (29.99 second). At the end of the search, the maximum objective was 8679 sec (144.65 mins), which represents the total travel time for all vehicles. The search finished with the optimal solution that had an objective value of 7349 sec (122 mins). Interestingly, this objective value was the minimum among all generated solutions indicating that the system has successfully found a global optimum solution.



In emergency cases, the system was examined after generating the routes for each bus. The drop-down button was used along the route to update the students who are currently in the route. Assuming that route number 1 has an emergency, case and needs the nearest hospital, the information was specified from the driver using the website application as seen below in Figure. Afterwards the nearest hospital was printed on the screen and directions were given on a map, simultaneously the system will reoptimize the remaining students, with the emergency location outputted as the starting point. After the emergency case has subsided the driver succeeds to either picking up or dropping off the remaining students on that specific route by following the map directions resulting from the re-optimization action.


 

CONCLUSIONS

After completing all the required aspects of this project, we accomplished implementing an Intelligent system for route optimization. That helps in improving delivery time, reducing fuel costs, and increasing customer satisfaction. As opposed to manual and inefficient scheduling of school bus routes locally, causing high costs and lengthy traveling time of vehicles. In addition, to poor emergency management when school buses are en route.

By conducting the literature review, it made it easy to pinpoint the weak points and strong aspects of current search algorithms used in route optimization and how they are incorporating fundamental machine learning models. Then by adopting the Pughs method, we were able to build the project’s mainframe, which consists of four main segments listed according to importance; optimization and emergency algorithm, microcontroller and GPS, database, and user interface. The hardest part was in making sure all segments of this project we're working coherently and responsive.

The tested intelligent system showcased that it can truly interact with the user and output great results depending on the demand. In turn, our objectives of the project were met, which are to develop an AI-based route optimization algorithm for school transportation that reduces the travel time delay. Furthermore, provide dynamic route optimization that adapts to emergency conditions relating to the driver, students, and bus. The Simulated Annealing (SA) algorithm was utilized with the assistance of Google Maps APIs and Firebase Google Cloud to deliver a web application for school bus providers. The output will be displayed through an LCD screen that is connected to a Raspberry Pi board, which will run the route optimization algorithm.

There is always room for improvement, especially in the field of optimization solutions. It is well known in optimization problems that till now, there isn’t a solution that is 100% optimal. For that matter, we propose the following improvements:

  • Adopt a hybrid of search algorithms to obtain better optimality.

  • Increase the system's intelligence by adding a machine learning model to learn which student are always late to enter the bus

  • Replace any hardware with software material.

  • Swap the Raspberry Pi board with Google’s virtual supercomputer.

  • Swap the GPS module by taking readings from mobile devices’ GPS module via a mobile application.

As a team of four engineering students, we have gained much knowledge not only about the science of optimization but rather how to implement it in a real-world scenario and market it to the community. We believe that by completing this project, we have fulfilled all the requirements to take what we have learned from our college years and apply it when we are employed or continuing our studies.



 

Team Member



Project Advisor

Dr. Mohammad Awedh

Project Co-Advisor

Dr. Fatima Abdelhedi

SDP Committee Coordinator

Dr. Thangam Palaniswamy



Recent Posts
Archive
Search By Tags
bottom of page