Release 2018a Highlights

It's March again, and you know what that means. The Earth churns along in space on its tilted axis approaching its vernal equinox, while MathWorks churns out another release. Amongst many other things, release 2018a continues to improve on deep learning and data analytics in MATLAB, automotive-related features in both MATLAB and Simulink, and provides a new dedicated tool for predictive maintenance.

18a improves on every part of the deep learning workflow. The Image Labeler app lets you automate pixel labeling for semantic segmentation. You can create better, more refined networks using the new support package for Keras TensorFlow, along with additions and improvements to the training process. If you're using any of these more sophisticated models, you can use GPU Coder to generate CUDA^® code and run them directly on video GPUs. You can also deploy trained models to Intel and ARM processors.

18a also improves the data analytics workflow. There are new preprocessing capabilities for tables and timetables, like being able to retime and synchronize data collected at different rates. You now have access to even more functions that just work with big data. If you're doing text analytics, you can quickly extract HTML text and identify common phrases and proper nouns.

One application of data analytics is predictive maintenance. If you're a frequent flier, chances are you've been on a flight delayed due to maintenance issues. Instead of having to fix something after it breaks or doing unnecessary checkups, predictive maintenance can detect an issue before something fails and help schedule repair time that doesn't delay your flight. The new Predictive Maintenance Toolbox helps to monitor machine health, estimate time of failure, and manage all the machine data. But perhaps more importantly, it helps you get started. It provides built-in examples so you can leverage and customize. So just check out the product documentation to get going.

18a features many new and improved tools for automotive applications. The new Vehicle Dynamics Blockset lets you evaluate a vehicle's ride and handling characteristics before conducting necessary yet expensive and possibly dangerous testing in the real world. It comes with a sweet 3D environment to visualize simulations, which can also pipe data back into Simulink. You can immediately get started with preassembled vehicle dynamics models and maneuvers, or customize them for your specific needs.

Speaking of maneuvers and responses, those of you working on 8s and automated driving could utilize the new Driving Scenario Designer to interactively define actors and driving scenarios for testing control and sensor fusion algorithms. And if you are designing adaptive cruise control and lane-keeping assist algorithms, you can use prebuilt blocks instead of having to start from scratch.

MATLAB's Model-Based Calibration Toolbox™ integrates more closely with Simulink. For example, using just a few clicks, you can fit models to your engine test data and export them directly into Powertrain Blockset™ models. And finally, 18a enhances vehicle connectivity with CAN FD and XCP over Ethernet support in both MATLAB and Simulink, enabling Model-Based Design and calibration workflows. Be sure to check out the rest of what's new while you download and install the latest release. Thanks for watching. And don't hesitate to send us your feedback.