In the ever-evolving world of engineering, the integration of Artificial Intelligence (AI) with model-based design is creating significant advancements, particularly in the automotive industry. This blog post covers the intricate world of AI for engineering leaders, specifically focusing on model-based design, highlighting its benefits, challenges, and real-world applications.

Model-Based Design is the systematic use of models throughout the development process that improves how you deliver complex systems. You can use Model-Based Design with MATLAB and Simulink to shorten development cycles and reduce your development time by 50% or more. Source

Understanding Model-Based Design

Model-based design (MBD) is a method that uses models as the primary means of information exchange between various stages of product development. This approach allows engineers to simulate, test, and validate their designs before physical prototypes are created. The integration of AI with MBD enhances the ability to predict and improve system performance, leading to more efficient and innovative solutions.

Physical Systems and Model-Based Design: Understanding the Fundamentals

A physical system, whether found in nature, on a test bench, or within an application, typically comprises a set of interconnected and interacting objects or components designed to perform specific tasks or functions. In the context of model-based design, studying a physical system involves analysing the internal mechanisms using fundamental physical laws and engineering principles.

Within a physical system simulation environment, a model represents a selected object or component as an approximation. This means that models can vary in their levels of abstraction, depending on how closely they need to match the real-world physical component they represent. The fidelity of a modelโ€”how accurately it reflects the actual componentโ€”depends on the simulation objectives for the current design phase and the available information used to characterise and verify the model. This information can include manufacturer data sheets, measurements, and calculated parameters from targeted performance characteristics.

Model-based design enables a design flow described by the V-diagram
Model-based design enables a design flow described by the V-diagram (Source)

Model-based design enables a design flow described by the V-diagram. At each stage of the process, models of appropriate abstraction for that phase are used to represent system components and to simulate the design. These models can initially be high abstraction, modelling macro behaviour for the purpose of general concept and conceptual architecture exploration. Under different design phases where more specific questions are being asked, the models can become more detailed as the design under study becomes more finalised until the entire system is rendered in models of appropriate accuracy to validate the design for production.

The Automotive Industry: A Leader in AI and MBD Integration

The automotive industry stands at the forefront of integrating AI with MBD. One notable application is in the development of advanced driver-assistance systems (ADAS) and autonomous driving technologies. These systems require the integration of various sensor data and control algorithms to make real-time decisions. For example, pedestrian detection systems use AI to analyse video or image data and make quick decisions to avoid collisions.

Benefits of Model-Based Design

With design engineers under pressure to achieve better quality designs that can be brought to market faster and at an increasingly lower cost of manufacture, the need to confirm the robustness of a design now extends beyond the verification of nominal system performance and into such areas as the impact of component tolerances, yield, and manufacturing options. The same criterion applies to existing design platforms which are often reused with updated or new functionality or improved performance features. To support this, it is essential to leverage model-based design and simulation technology to facilitate the production of robust, tolerance-neutral systems for which the yield can be determined, sensitivity and stress analyses undertaken, and critical parts identified prior to physical prototyping. Here are some key benefits of a model-based design:

  1. Reduces Expensive Hardware Iterations: Designers can verify the design of a system and optimise the hardware and software implementations before its construction or do so with existing designs and avoid potentially expensive physical prototype iterations.
  2. Enhances Functional Safety: Virtual prototyping allows for aggressive and extensive testing, ensuring safety and efficiency. This includes detecting potential embedded software failures due to erroneous sensor signals, voltage drops, or other fault circumstances, providing sanity checks for intermittent steering stability scenarios.
  3. Time-to-Market Reduction: MBD methodology can significantly reduce time-to-market for new designs, providing critical flexibility to meet customer performance requirements.
  4. Warranty Overhead Cost Reduction: Simulation on a system with component tolerances allows for invaluable statistical analyses and worst-case analysis (WCA) performance evaluations. This enables designers to predict manufacturing yield, estimate final quality assurance inspection pass rates, and anticipate potential field failures. WCA simulation helps avoid expensive warranty callbacks by enabling early design change decisions.

Advantages of Model-Based Design

  1. System-Level Simulation: MBD allows for the simulation of the entire system, incorporating various components and their interactions. This holistic view helps in identifying potential issues early in the design phase.
  2. Cost and Time Efficiency: By simulating and validating designs virtually, MBD reduces the need for multiple physical prototypes, saving both time and resources.
  3. Enhanced Accuracy: The integration of AI with MBD improves the accuracy of simulations. AI models can process large datasets and identify patterns that traditional methods might miss, leading to more reliable and precise designs.

Challenges in Implementing Model-Based Design

Despite its advantages, implementing MBD comes with its own set of challenges:

  1. Interdisciplinary Collaboration: Engineers from various domains need to work together seamlessly. This requires effective communication and understanding of different technical languages.
  2. Skill Gaps: Engineers might need additional training to effectively use AI and MBD tools. This necessitates upskilling and continuous learning.
  3. Integration with Existing Tools: Many engineering teams use a variety of tools. Ensuring interoperability between these tools and MBD platforms is crucial for smooth workflow.

Overcoming Challenges: The MathWorks Approach

MathWorks addresses these challenges through various strategies:

  1. Consulting Services: They provide consulting services to help companies implement MBD and AI effectively. This includes developing implementation plans, conducting pilot projects, and providing ongoing support.
  2. Training and Upskilling: MathWorks offers training programs and on-ramp courses to help engineers and managers understand and utilise their tools. These courses are designed to be accessible and practical, providing hands-on experience with AI and MBD.
  3. Technical Support: MathWorks provides extensive technical support, ensuring that users can troubleshoot and solve problems efficiently.

Real-World Applications & Success Stories

NXP Success Story

One compelling success story comes from NXP, a leading semiconductor manufacturer. NXP leveraged MathWorks tools to integrate AI with signal processing, improving the performance of their products. By using a simple feed-forward neural network, they enhanced the accuracy of their analog-to-digital converter (ADC) systems, demonstrating the practical benefits of combining AI with MBD.

Post-Correcting ADC Errors with Neural Networks
Post-Correcting ADC Errors with Neural Networks (Source)

ASML Success Story

Schmitt-Weaver used MATLAB, Statistics and Machine Learning Toolbox, and Deep Learning Toolboxโ„ข to develop a method for generating virtual metrology.

ASML Develops Virtual Metrology Technology for Semiconductor Manufacturing with Machine Learning (Source)

Schmitt-Weaver used the Neural Network Time Series Prediction and Modeling app to learn how to prepare data for use with Deep Learning Toolbox. Using the app, he generated and exported the example code, which gave him a more detailed understanding of how the functions could be used together. As his competence increased, he was able to build on the generated code using examples from the vast multidisciplinary user community on MATLAB Central.

Mercedes-Benz Simulates Hardware Sensors with Deep Neural Networks

Automated workflow for deploying virtual sensors to powertrain ECU.
Automated workflow for deploying virtual sensors to powertrain ECU.

Mercedes-Benz recently used MATLABยฎ and Simulinkยฎ to establish a new workflow for deploying virtual sensors, such as those that simulate the functionality of a piston pressure sensor. These sensors are based on deep learning networks designed to run on resource-limited ECU microcontrollers. This automated workflow replaced a manual workflow that was both slower and relied on a trial-and-error approach.

โ€œThe MathWorks team helped us develop an easy-to-use pipeline for creating and integrating neural networks into our vehicle controller units,โ€ says Katja Deuschl, AI developer at Mercedes-Benz. โ€œWith this pipeline, we are now able to create and deploy different types of neural networks for virtual sensors and a variety of other applications.โ€

The Role of MathWorks in AI & MBD

MathWorks, a leader in mathematical computing software, plays a crucial role in facilitating the integration of AI with MBD. Their tools like MATLAB and Simulink provide a platform for engineers to develop, simulate, and deploy AI-enhanced models. Engineers can design systems incorporating various domains such as signal processing, mechanics, and control systems, enabling a comprehensive view of the entire system.

The Future of AI and Model-Based Design

The future of AI and MBD looks promising, with potential applications expanding across various industries. As AI technology continues to evolve, its integration with MBD will likely lead to even more innovative and efficient solutions. Companies that embrace this integration will be better positioned to develop cutting-edge products and maintain a competitive edge.

Closing Remarks

AI and model-based design represent a powerful combination that is transforming engineering and product development. While challenges exist, the benefits far outweigh them, making this integration a critical area of focus for forward-thinking engineering leaders. MathWorks, with its robust tools and support systems, stands as a pivotal partner in this technological evolution, helping companies navigate and harness the full potential of AI and MBD.

For those interested in exploring this further, reaching out to MathWorks for consultation and training can provide valuable insights and a practical path forward. As the landscape of technology continues to shift, staying ahead with AI and MBD will be key to driving innovation and achieving success in the engineering domain.