Editor’s Pick: Engineering Tools Feature Artificial Intelligence, Robotics and More
MathWorks Release 2019b update adds six new deep learning models and a boost to autonomous automotive engineering capabilities.
October 2, 2019
Dear DE Reader,
Hockey legend Wayne Gretzky once said the secret to his success was, “I skate to where the puck is going to be, not to where it has been.” When the editorial team ‘skates’ to where things are happening in engineering technology, we usually find MathWorks waiting there for us. There is a new release of MATLAB and Simulink capabilities in support of artificial intelligence (AI), robotics, automotive engineering and more tools for engineering success. It was a no-brainer to select MathWorks Release 2019b as our Editor’s Pick of the Week.
Our readership surveys reveal strong and growing interest in AI and deep learning (DL) for product design. MATLAB is there in the 2019b update with various new features, including new tools for data preparation and labeling, with a focus on identifying and visualizing what DL experts call ground-truth data. It is the process of gathering the right provable data for training the neural net, as the foundation for helping your AI model predict or seek the right data. Think of it as establishing and understanding the gold standard for your project.
Smart products generally have the internal equivalent of a network architecture as a base layer for building the digital side of the product. In the 2019b update, there are new capabilities for building advanced networks, including a Deep Network Designer app, to graphically design and analyze a deep network and to generate MATLAB code required to build it.
We’ve known since the earliest days of CAD that interoperability is important in engineering software, and it turns out it is also important in the new world of AI and DL for product design. Much of the work to date in AI is based on open source data models, modified by users as needed for their products. This update adds six new deep learning models, making it even more useful when building CUDA code for a deep learning project.
Autonomous automotive engineering gets a boost in this new update. MathWorks has added support for Unreal Engine sensor models. This allows the user to integrate a Simulink model with a camera, lidar or radar sensor model running in an Unreal Engine scene.
MathWorks is known for its use in robotics engineering. In addition to new features in the Robotics System Toolbox, R2019b introduces two products. Navigation Toolbox is for the design, simulation and deployment of algorithms for planning and navigation. ROS Toolbox is also for design, simulation and deployment but is specific to the Robot Operating System (ROS). This new toolbox gives engineers an interface between MATLAB, Simulink and ROS and ROS2, for generating embedded system software for ROS nodes.
Among the MATLAB highlights in R2019b is the introduction of Live Editor Tasks. This helps users interactively explore parameters, preprocess data and generate MATLAB code that will become part of a live script. This means MATLAB users can focus on the task instead of the syntax or complex code, and they can automatically run generated code to quickly iterate on parameters through visualization.
If you use any of the tools in the MathWorks suite, the 2019b update means Christmas came early. Be sure to read this week’s Product Brief for more details.
Thanks for reading; see you next week with another Editor’s Pick of the Week.