R2020b also introduces new products that build on artificial intelligence (AI) capabilities, speed up autonomous systems development, and accelerate creation of 3D scenes for automated driving simulation.
Among the hundreds of new and updated features, MATLAB adds new bubble and swarm charts, the ability to diff and merge App Designer apps with the MATLAB Comparison Tool, and customisable figure icons and components to MATLAB apps. Also, in addition to Simulink Online to view, edit, and simulate Simulink models through web browsers, R2020b adds the ability to generate code up to 2X faster for referenced model hierarchies in Simulink and includes new automerge functionality that helps automate continuous integration workflows.
Release 2020b highlights include:
- AI and Deep Learning
The new Deep Learning HDL Toolbox provides algorithm developers and hardware designers with functions and tools to prototype and implement deep learning networks on FPGAs and SoCs. It provides prebuilt bitstreams for running a variety of deep learning networks on supported Xilinx and Intel FPGA and SoC devices. Profiling and estimation tools help to customise a deep learning network by exploring design, performance, and resource utilisation trade-offs.
By using Deep Learning HDL Toolbox, engineers will be able to customise the hardware implementation of their deep learning network and generate portable, synthesizable Verilog and VHDL code for deployment on any FPGA with HDL Coder and Simulink.
Starting in R2020b, Deep Learning Toolbox supports Simulink with Image Classification and Network Prediction blocks to help simulate and generate code for deep learning models. Updates to the Experiment Manager app support training of multiple deep learning networks in parallel and tuning of hyperparameters using Bayesian optimisation. The Deep Network Designer app can now train networks for image classification, semantic segmentation, multiple-input, out-of-memory, image-to-image regression, and other workflows.
- Autonomous Systems
Two new products and a major update for autonomous systems are included in R2020b. Lidar Toolbox is a new product that provides algorithms, functions, and apps for designing, analysing, and testing lidar processing systems. UAV Toolbox is another new product that provides tools and reference applications for designing, simulating, testing, and deploying unmanned aerial vehicle and drone applications. Updates to Sensor Fusion and Tracking Toolbox include the ability to track objects using the grid-based random finite set (RFS) tracker and the ability to create trajectories using earth-centred waypoints.
- Automotive
For the automotive industry, the new RoadRunner Scene Builder product, part of the RoadRunner product family, automates the creation of road networks from HD Maps. AUTOSAR Blockset updates enable the use of schema version 4.4 for import and export of ARXML files and generation of AUTOSAR-compliant C code.
It also offers Linux executables for adaptive models, helping to create an AUTOSAR adaptive executable to run as a standalone application. Vehicle Dynamics Blockset now includes the ability to implement 6DOF trailers and vehicles with three axles and Simulink 3D blocks that offer the ability to visualize tractors and trailers in the Unreal Engine 3D environment.
