AI-driven Public Urban Transport Optimization: Introduction & Architecture
This blog post is part of a series of technical enablement sessions on SAP BTP for Industries. Check the full calendar here to watch the recordings of past sessions and register for the upcoming ones! The replay of the session related to this blog post and all the other sessions is available here.
Authors: Cesare Calabria and myself
We are trying to use the SAP Business Technology Platform in order address the challenges in the Urban Public Transport industry.
We start with an introduction to the Urban Public Transport, discuss its challenges and pain points, focusing on the big data management, and the solutions that we can offer to tackle them.
Afterwards we would see the architecture of the solution using services on top of the SAP Business Technology Platform.
These demos will give you a hands-on experience of the drafted prototype. You will learn not only how to design but also how to bring a solution into a proof-of-concept and how it can help businesses in the urban public transport industry to improve customer experience.
This doesn’t aim to be a Master Class on all of the pieces and parts of a solution, but rather highlighting how to go from the analysis of challenges for a specific industry to the implementation of a solution, through the design of good cloud native architecture.
We hope to get you inspired!
What is the Urban Public Transport Industry?
The urban public transportation industry refers to the sector involved in moving people from one location to another. In terms of people transportation, it includes various modes such as buses, trains, subways, and others. The industry’s primary goal is to provide safe, efficient, and reliable transportation services.
Here are some key facts about the urbanization trends world-wide.
Overall, more people in the world live in urban than in rural setting since 2010. In 2020, 56.2 percent of the world population was urban. Around 75% of the EU citizens are lining in the cities.
The traffic congestion and inefficient transport systems still persist, and, for example, they account for 24% of GHG emissions in the European cities. This is due to high population density of our cities that is also destinated to increase, since the movement from rural areas to cities never stopped.
The urban transport has impact to the society and the environment. For example, in the US public transportation is having 84% less carbon emission compared to by using regular cars. The electric and hybrid buses count is constantly growing and now around 80% if then is leading to clean technology.
The overall consumption of the gasoline is reduced by 6 billion gallon each year by using the public transport. Public transportation continues to be one of the safest modes of travel. Safe travel is a high priority of public transportation systems, state, and local governments.
The industry’s primary goal is to provide safe, efficient, and reliable transportation services.
The public transport systems are far from being efficient and well planned, resulting in several issues that we experience every day. Let’s see, what are the major challenges in the industry:
- Long detouring distance
- Improper departure frequency setting
- Crowded travel during rush hours
- Poor passenger comfort in vehicles
City expansion issues:
- Unreasonable layout of public transport lines
- Insufficient public transport infrastructures
- Insufficient number of stations
- Long waiting times
The main challenge for the transportation industry is to continuously improve the transportation service in the city, capture each anomaly and react in the real-time.
The only possibility to deal with so many and complex data is to drop the classic approaches and turn to AI.
How Can You Help, as an SAP Partner? The Proposed Solution
We are working with a data for one particular city, namely London in the United Kingdom. The data foundation is provided by external data services, where the bus location, timetables, road disruption are collected.
The solution would have two kinds of output. Namely the monitoring dashboards, which would help to realize a real-time analysis on the bus network, and it would enable to identity the issues early.
The second output would be the timetable optimalizations. Where the main benefit is the optimal path and frequency, together with shorter waiting time.
Four personas are involved in the scenario:
1. The first persona is John, a data engineer.
John would start the process with data ingestion and orchestration, where all the external data in collected with SAP Data Intelligence and it is ending in Data Lake. John is also responsible for the data modeling process, which is happening in SAP Datasphere.
2. The second persona is Emily, a transport analyst.
The data exposed in the Analytical Models, adjusted by the AI algorithm is visualized and monitored by Emily, the transport analyst. She has a ton of data, based on that she can analyze the current happening in the city.
3. The third persona is Tom, a data scientist and application developer.
In order to use AI, Tom would develop and deploy some algorithm in SAP AI Core. He would be responsible for algorithm training and maintenance withing the SAP AI Launchpad.
4. The fourth persona is George, a transport planner.
George is constantly working with the machine learning algorithm to test and train them. And make final decision on the transport improvements based on the data what the AI provides.
The Proposed Solution Architecture
Now, let’s see how we can design a solution architecture foundation to cover this business scenario.
In the central point we would find the SAP Datasphere, the next generation of SAP Data Warehouse Cloud. This is the place where all data is concentrated.
The SAP Data Intelligence is used for collecting the from the external sources, which includes the social media as well
The pipelines in the SAP Data Intelligence ingests the data to the Data Lake, which we are using as our main storage for unstructured data.
The vehicle use in the transportation companies are managed by the SAP Digital Vehicle Hub, which contributed to the final data model as well with master and transaction data related to the vehicle management.
The visualization and the story is deployed in the SAP Analytics Cloud, which is exposed in the SAP Build Work Zone, Standard Edition as well. The Work Zone is the final end-user interface, where the role-based access helps to assign the right access for each user.
The SAP AI Core and SAP AI Launchpad contributes to the collected data with machine learning algorithms.
Exploring the Possibilities
Let’s now explore some other ways we could have thought to build this architecture and the rationale behind the decision to choose the selected components and technologies.
Option 1 – with Twitter and Event Mesh: The Twitter might be used to collect the post published in the social media, especially the post about the incident, event, or road disruption, which had been published by users based on the real happening in the city. Twitter is supporting a webhook technology, which might be used to trigger a data event directly to the Event Mesh, which can distribute the data in the real rime. With that we might slightly introduce to the event driver architecture and there is no need to collect them by time to time. We would receive the data as soon as they are created.
Option 2 – with Event Mesh combined to other third-party resource: Currently the data from the UK Government is collected in a certain time interval. In the current option we would replace a pipeline in the SAP Data Intelligence to collect the data and we would push the data directly from the data source if such feature is supported by the data provider party. We saw the event driven architecture in the previous option and this might be implemented to other data providers as soon as they are supporting the webhook technology as well.
Option 3 – with SAP Integration Suite: The SAP Data Intelligence might be replaced with SAP Integration Suite. Whereas in the Cloud Integration the data ingestion would be developed and might be executed on the interval defined in the timer.
Option 5 – without SAP Build Work Zone: In the current option we are again touching the powerful CAP framework, but in our case, it would be used to create the end user interface using the SAP Fiori technology. This might be beneficial if you would like to reduce the costs for having the SAP Build Work Zone for the whole landscape. But on the other site it’s increasing the development effort to create the UI instead of leveraging the SAP Build Work Zone, which is designed for such content exposure for the end users. Other reason for that might be the overall architecture model if you would like to bind multi-tenant Fiori application in the SAP Build Work Zone, where the functionality is still under the investigation by SAP. The custom Fiori Launchpad created with single application router can address this challenge as well.
Option 6 – without SAP AI Core: The SAP AI Core can be replaced with SAP BTP, Kyma runtime. We can build machine learning application based on SAP BTP, Kyma runtime and integrate with streamline which is open-source Python library to create custom web apps for machine learning and data science.
Before going deep into the implementation, let’s watch a demo video of the solution:
Implementation Deep Dive
Please refer to the blog created by my colleague Cesare Calabria :