The Predictive Maintenance Toolbox from MathWorks operates with sensor and failure data (from Simulink models, for example), which users can analyze and label precisely. Various methods, models and an extensive pool of reference examples are available in this connection. By these means, a company can develop and test its own algorithms for predictive maintenance at reasonable cost. The toolbox is available to try out in a free 30-day test version .

When it comes to deep learning, embedded vision and autonomous systems, too, MathWorks is keeping its eye on the ball: Matlab for Deep Learning is being short-circuited with TensorRT accelerators via GPU Coder . This is intended for complex AI applications with a high throughput and low latency. Internal benchmarks have shown that, in this way, results that are five times faster can be achieved with Nvidia GPUs compared to TensorFlow. The system is due to be unveiled at the GPU Technology Conference in October 2018.