As robot systems become increasingly complex, engineers are using “models” to replace documents

Why does complex system development always “derail” at the prototype stage? Perhaps the answer is not to work harder on writing documents, but to adopt a different engineering approach.

MBSE（Model-Based Systems Engineering）

In many people’s imagination, robot development should look like this: engineers design structures, programmers write algorithms, testers run simulations—everything proceeds step by step.

But the reality is often completely different.

The mechanical team updates the sensor mounting position but fails to communicate it to the algorithm team; the algorithm team modifies the obstacle avoidance logic, but the scheduling system is unaware of the change; the update rhythm of the warehouse management system (WMS) conflicts with the robot scheduling rules.

What is the result?

By the time the project reaches the prototype stage, problems are discovered—over 80 logic conflicts.

Such situations are not uncommon in complex system development.

Now, more and more companies are beginning to adopt a new method to solve this problem: MBSE (Model-Based Systems Engineering).

 From “Document-Driven” to “Model-Driven”

Traditional systems engineering development typically relies on a large number of documents: requirements specifications, design documents, interface protocols, test documents... Each team has its own version of the files. The problem lies in the lag in document updates, the ambiguity of natural language, and the continuous “distortion” of information during transmission. This ultimately creates a typical phenomenon: every team believes they are correct, but the system simply cannot run. MBSE attempts to change all of this. Its core concept is very simple: use formalized digital models to replace traditional documents as the core of development. Under the MBSE framework, the entire process of system development—requirements analysis, conceptual design, detailed design, simulation verification, and physical implementation—revolves around a unified digital model. The model becomes the “single source of truth” for the entire system. Engineering teams no longer rely on passing documents back and forth; instead, they collaborate based on the same model.

How MBSE Solves Real Engineering Problems

Take the development of an intelligent logistics robot as an example.

Logistics robots may seem simple, but they involve multiple complex systems behind the scenes: - LiDAR perception,

- path planning algorithms,

- multi-robot collaborative scheduling,

- warehouse management system (WMS).

The most critical among these is the dynamic path planning and multi-robot collaborative scheduling module. This module directly determines:

- operational efficiency,

- operational stability,

- warehouse throughput capacity.

At the same time, it is also the area where traditional development models encounter the most problems. In one real project, the team discovered during the prototype phase that:

- the path obstacle avoidance logic did not match the sensor parameters,

- the scheduling rules conflicted with the WMS inventory update rhythm,

- the multi-robot task allocation logic was inconsistent.

In this single module alone, over 80 logical conflicts emerged.

The project nearly came to a standstill.

Using SysML to “Draw Out” the System

To solve the problem, the team introduced the MBSE approach and used the SysML modeling language to construct a system model. Following the classic V-model process of MBSE, the team completed a full model-based development.

Figure 1: MBSE V-Model Process

01

Requirements Modeling

Through SysML requirement diagrams, the team transformed vague requirements into 126 quantifiable requirements, such as:

- Dynamic obstacle avoidance response time ≤ 0.5s

- Support for ≥ 50 robots operating in parallel

At the same time, traceability relationships between requirements, design elements, and test indicators were established. When requirements change, the scope of impact can be quickly identified.

02

System Behavior Modeling

The team used various SysML models to describe the system’s operational logic:

- Activity diagram: describes the robot path planning process

- Sequence diagram: depicts the information interaction between the robot, the dispatch center, and the WMS

- State machine diagram: defines the state transitions of the robot under different working conditions, including:

  - Normal driving

  - Dynamic obstacle avoidance

  - Fault alarm

These models allow different specialized teams to intuitively understand the system logic.

 03

Executable Simulation

Next, the SysML model was transformed into an executable simulation model.

The team accessed the three-dimensional environmental data of the real warehouse and simulated more than 20 typical operating conditions for simulation verification. Potential problems were identified and optimized during the design phase.

This truly achieved: shifting issues forward, resolving them at the design stage.

Real Changes Brought by MBSE

After applying MBSE to the core module, the results were very significant. For this module alone: the R&D cycle was shortened by 40%. The entire project also underwent changes:

- Cross-team collaboration efficiency was significantly improved

- Requirements change response time was reduced from 2 days to 4 hours

- Simulation verification accelerated design iteration

More importantly, the model can be continuously reused. When the warehouse scale is upgraded, the system does not need to be redesigned; it only needs to be expanded on the basis of the existing model.

Two New Trends in MBSE

With the development of artificial intelligence, digital twins, and cloud computing, new technical directions are emerging in MBSE.

01

AI-Enhanced Simulation

Machine learning is being embedded into MBSE models. For example, in spacecraft thermal control design, NASA has introduced reinforcement learning into MBSE simulations. AI can automatically search for the optimal parameter combination. The results are:

- Simulation efficiency improved by 40%

- Predictive analysis of spacecraft in-orbit thermal control status achieved

02

Digital Twin Integration

MBSE models are beginning to connect with real systems in real time. In the field of intelligent manufacturing, Siemens synchronizes its production line MBSE model with equipment IoT data. The virtual model can map the real production line status in real time. The system can even predict equipment failures in advance. The failure prediction accuracy rate reaches 92%. This has significantly reduced unplanned downtime on the production line.

Of course, the promotion of MBSE in industry still faces challenges.

The first is the talent issue. MBSE engineers not only need to understand systems engineering, but also need to master standardized modeling methods. Such composite talents are currently relatively scarce.

The second is tool cost. The licensing fees for mainstream commercial MBSE tools are high, making them often unaffordable for small and medium-sized enterprises.

However, the situation is improving.

First, open-source tools are developing rapidly, such as Eclipse Capella and Papyrus for SysML. These tools have significantly lowered the application threshold for MBSE.

At the same time, an increasing number of universities are beginning to offer MBSE courses in systems engineering and intelligent manufacturing programs to cultivate composite talents, and the industry’s talent pool is gradually growing.

An Emerging New Engineering Paradigm

Industry research shows that over 40% of enterprises are transitioning to model-driven engineering. In complex system domains such as aerospace, the application rate of MBSE has exceeded 50%.

This means that MBSE is becoming the mainstream approach for developing complex systems.

In the era of the Internet of Everything, systems are becoming increasingly complex:

- Smart equipment

- Intelligent manufacturing production lines

- Autonomous driving systems

- Urban transportation networks

These systems all exhibit distinct cross-domain coupling characteristics. Traditional document-driven methods are no longer sufficient to handle these systems.

MBSE provides a new engineering paradigm: using models to connect requirements, design, simulation, and the real world.

From logistics robots in smart factories, to aerospace vehicles, and to future urban transportation systems, MBSE is enabling modeling and simulation technologies to truly step out of the laboratory and into industrial sites.

Moreover, MBSE is gradually becoming an important force driving the development of high-end manufacturing.