MathWorks R2020b Releases MATLAB and Simulink Bring Hundreds of New Features

MathWorks has launched the R2020b version of the MATLAB and Simulink product series. New features in MATLAB make it easier for users to process graphics and create apps, while updates to Simulink focus on helping users achieve faster and more convenient access. With the newly launched Simulink Online, users can use Simulink directly through a web browser. R2020b also launched new products based on artificial intelligence (AI) to accelerate the development of autonomous systems, quickly create 3D simulation scenarios for autonomous driving, and more.

Among hundreds of new and upgraded features, MATLAB has added bubble charts and clustered scatter plots; provides the ability to compare and merge apps created with App Designer using the MATLAB comparison tool; and features such as adding customized icons and components to your MATLAB App. Simulink has launched Simulink Online, which allows users to view, edit, and simulate Simulink models in a web browser; improves the code generation capability of the reference model architecture (reference model) by up to 2 times; and the new automatic merge function helps users achieve automated continuous integration workflows.

Other highlights of R2020b:

AI and Deep Learning

The new Deep Learning HDL Toolbox provides algorithm developers and hardware designers with functions and tools to create prototypes and implement deep learning networks on FPGAs and SoCs, including preset bitstreams to support running various deep learning networks on Xilinx and Intel FPGA and SoC devices; feature analysis and evaluation tools to help customize deep learning networks by weighing design, performance, and resource utilization. Using Deep Learning HDL Toolbox, engineers can customize the hardware implementation of deep learning networks and use HDL Coder and Simulink to generate portable, synthesizable Verilog and VHDL code and deploy it on any FPGA.

Starting with R2020b, Deep Learning Toolbox supports Simulink for simulation and code generation of deep learning models with image classification and network prediction modules. The Experiment Manager App has been upgraded to support parallel training of multiple deep learning networks and hyperparameter tuning using Bayesian optimization methods. The Deep Network Designer App can now train networks to implement workflows such as image classification, semantic segmentation, multiple inputs, memory overflow, and image-to-image regression.

Autonomous Systems

R2020b introduces two new products for autonomous systems and one major update. Lidar Toolbox is a new product that provides algorithms, functions, and apps for designing, analyzing, and testing lidar processing systems. UAV Toolbox is another new product that provides tools and reference applications for designing, simulating, testing, and deploying drone applications. Updates to Sensor Fusion and Tracking Toolbox include the ability to track objects using a grid-based random finite set (RFS) tracker and create trajectories using waypoints in a geocentric coordinate system.

car

For the automotive industry, the new RoadRunner Scene Builder product in the RoadRunner product family can automatically create road networks using high-precision maps. Updates to AUTOSAR Blockset enable import and export of ARXML files and generation of AUTOSAR-compliant C code according to version 4.4 of the AUTOSAR standard. It also provides Linux executables for adaptive models to help create adaptive AUTOSAR executables that run as stand-alone applications. Vehicle Dynamics Blockset can now implement 6DOF trailers and vehicles with three axes and simulate Simulink 3D blocks, so that tractors and trailers can be visualized in the Unreal Engine 3D environment.