Where is the business logic in digital twin-based fleet management solutions?

A fleet management solution comprises two component groups. The first group focuses on asset management, driver behavior, usage reports, general reporting, analytics, and tracking of the current and historical positions and states of vehicles. These aspects provide valuable information to fleet management personnel, enabling them to consume the fleet’s data comprehensively.

The second component is operational, providing contextual information about the fleet’s ongoing task execution. It establishes connections between entities within the fleet domain, such as drivers, vehicles, current trip plans, destinations, weather conditions, and traffic situations. This contextual information is essential for fleet management to make informed decisions. When the context encompasses all the necessary entities and integrates with external tools, it facilitates informed decision-making and enables automated decision-making.

We will delve deeper into the operational aspect of a fleet management solution that can significantly benefit from an implementation based on the digital twin concept. We have developed an end-to-end solution as part of our exploratory efforts and approach validation. The solution demonstrates data flow and connections between various components.

Digital twin key features

The digital twin concept offers numerous utilization options, and many solutions take advantage of a relevant subset of its features. In our implementation, we have incorporated the following key features:

  • Situational awareness
  • System-of-systems
  • Simulation

Situational awareness relies on a digital twin model that accurately represents entities, properties, and relationships within the fleet domain. Twin modeling is elastic in respect of future updates. It is better to start with a small twin model that covers initial functionality and expand it in the future as needed. The twin model’s instance reflects real-world fleet objects and can continuously update with real-time data.

This comprehensive contextual information serves as the foundation for developing services on top of it. Situational awareness ensures reliable information for decision-making. It provides an equivalent experience to physically looking out of a window to count vehicles in a parking lot or obtaining such information from a dashboard. It is true even for complex details that may not be easily observable in the real world.

The digital twin also serves as an integrator, playing a crucial role in adopting a system-of-systems approach to gather data from various fleet management components and form the current state of the fleet. Once collected, data is ingested into digital twin properties or referenced to the information source. Each digital twin graph entity consolidates all the necessary information to support decision-making and downstream services.
We can leverage a digital twin model for simulations once we have a digital twin model. The first group of simulations involves testing business logic, decision logic, and automation implementation by simulating different fleet scenarios. It allows us to assess the effectiveness of our strategies in complex situations that are difficult to record and replicate in the real world.

The second group of simulations focuses on evaluating the feasibility of a fleet plan. These simulations consider predictions such as weather and traffic conditions and the probability of events such as vehicle failures. By considering these factors, the simulations assess the executability of the plan and calculate a risk score.

Business and decision logic component

Our solution’s digital twin-component represents the fleet’s current state but does not inherently contain business logic. Let’s consider a scenario where a vehicle is assigned to visit multiple destinations for customer maintenance work, with customers notified in advance about the scheduled visit time window.

The vehicle continuously sends its position and telemetry data, and a subset of this data goes through the IoT platform to the Digital Twin Graph. When the vehicle’s digital twin receives the position update, it triggers a monitoring service that calculates whether the current state aligns with the planned itinerary. All the necessary information for evaluation is either in the digital twin graph or pulled from specific services.

An alert is triggered if the estimated arrival time exceeds the scheduled time window. The decision logic can be complex, incorporating factors beyond comparing estimated arrival times. For instance, customer service levels may dictate different maximum waiting times based on customer status. A subject matter expert (SME) should manage such decision logic and may require periodic modifications to adapt to changing requirements.

DMN

To tackle decision complexity and provide a tool for subject matter experts (SMEs), we have utilized the DMN (Decision Model and Notation) modeling language. DMN offers a precise specification of business decisions and rules, making it understandable for various stakeholders involved in decision management. The graphical notation used in DMN is directly executable through an interpreter.

We have adopted a microservice approach in our implementation, leveraging the Kogito framework powered by Quarkus. It allows us to package decision diagrams into containers and execute them in a cluster as required. By utilizing this approach, we ensure scalability and flexibility in managing decision-related processing. DMN example:

DMN is stateless. One observation of a delay should not immediately trigger an alert. It is important to track the development of the delay and determine the appropriate course of action for a remedy process. To achieve this, we can incorporate another tool from the DMN family called BPMN (Business Process Model and Notation).

BPMN is a widely recognized graphical notation language specifically used for modeling processes. The Kogito framework is capable of executing BPMN diagrams as well. DMN and BPMN are complementary tools within the same family and can be of seamless use. DMN can be embedded within BPMN diagrams to provide decision logic. Moreover, the DMN family encompasses additional members, offering a broader range of tools based on specific use cases.

BPMN

By incorporating BPMN into our solution, we introduced a stateful component to handle delays. Upon initial observation of the delay, creating a BPMN instance allows us to continuously monitor the delay’s development. The customer receives a notification in case of postponement if there is consistent confirmation over certain occurrences.

This notification enables the customer to cancel the visit or wait for the delayed service. By utilizing BPMN in this manner, we allow a more proactive and customer-centric approach to managing delays within the fleet management process.

A BPMN diagram captures all these details:

To integrate the BPMN instance into our system, we store the BPMN instance ID as a property within the destination digital twin. It demonstrates the system-of-systems role of the digital twin, which facilitates the integration of the current fleet state with references to the corresponding business process and decision components.

By associating the BPMN instance ID with the destination digital twin, we establish a link between the real-time state of the fleet and the specific business process and decision-making aspects relevant to that destination.

Conclusion

This elegant implementation serves as a compelling demonstration of how a solution built upon a digital twin, enhanced with business process and decision logic, can significantly improve customer services while also opening up new avenues for service innovation. The flexibility of the digital twin allows for modeling various domains, enabling tailored solutions to meet specific industry needs.

We are committed to pushing the boundaries further by exploring more complex use cases and incorporating cutting-edge technologies into our solution. For instance, we are actively working on integrating generative AI to unleash even greater potential for optimizing fleet management operations.