How to Build an Autonomous Anything
Autonomous technology will touch nearly every part of our lives, changing the products we build and the way we do business. It’s not just in self-driving cars, robots, and drones; it’s in predictive engine maintenance, automated trading, medical image interpretation, and other applications. Autonomy—the ability of a system to learn to operate independently—requires three elements:
- Massive amounts of data and computing power
- A diverse set of algorithms, from communications and controls to vision and deep learning
- The flexibility to leverage both cloud and embedded devices to deploy the autonomous technology
In this talk, Jim Tung shows you how engineers and scientists are combining these elements, using MATLAB® and Simulink®, to build autonomous technology into their products and services today—to build their autonomous anything.
What's New in MATLAB and Simulink
Engineers and scientists worldwide rely on MATLAB® and Simulink® to accelerate the pace of discovery, innovation, development, and learning. In this presentation, you will see how new capabilities in the latest releases will help your research, design, and development workflows become more efficient. Learn how you can accelerate Model-Based Design through enhanced simulation performance in Simulink and use machine learning and deep learning in MATLAB for data analytics and computer vision applications.
Data Analytics and Technical Computing
Big Data and Machine Learning Using MATLAB
Predictive analytics is the engine of evidence-based decision-making. Today, big data and engineering techniques bring many opportunities to the world of analytics. However, using data to build accurate and robust models for prediction requires a combination of equipment, expertise, and statistical know-how.
In this session, Seth and Amit discuss strategies and techniques for handling large amounts of data. Attendees will learn about tools and algorithms used to create machine learning models that learn from data and scale those models up to big data problems.
- Accessing data in large text files, databases, or from the Hadoop Distributed File System (HDFS)
- Processing data that does not fit in memory
- Applying machine learning techniques to explore and prepare data for modeling
Automated Product Quality Inspection
Visual quality inspection of manufactured or processed products and commodities before packaging at the end of the processing line is a critical step for many industries. Shape defects, dimensional variations, and surface defects are some of the types of defects that should not pass a quality check. Consequences industries face in case of compromised quality inspection include:
- Negative impact on the brand value, which leads to loss of business
- High incurred costs when product must be recalled from the market
Currently, there are two solutions employed by the manufacturing and processing industries: manual and automated inspection. Automated inspection systems employ customized and specialized image processing techniques to perform quality check.
Manual inspection, subjective in nature, may compromise consistency. Manual inspection is error-prone due to fatigue and boredom. In case of dimensional checks, manual inspection is time consuming and cannot be performed for every product if the production rate is very high.
For automated systems reliant on customized image processing techniques, adaptability to a new variant or product is challenging. In some cases, depending on the extent of changes, automated systems may need to be developed again from scratch, which is both time consuming and expensive.
To ensure that only good products are delivered to customers, a side effect of automated systems may be that some good products may be rejected. This adds losses to the industry.
Integrating MATLAB Analytics into Enterprise Applications
MATLAB® applications and components can be deployed to a variety of platforms, providing you with the flexibility to determine the best solution for your organization. You can deploy any MATLAB program covering a range of industries and applications such as data analytics, semiconductor/electronics, manufacturing systems, image processing, aerospace and defense, and financial services. All applications and components are encrypted to protect your intellectual property and can be shared royalty free.
In this session, Pallavi presents various deployment options available in MATLAB to integrate your applications with today’s IT infrastructures without having to recode.
- Royalty-free distribution of applications to users who do not need MATLAB
- Integration of compiled MATLAB with C/Java/.NET/Python
- Large-scale deployment to enterprise systems
- Deployment of MATLAB code against Hadoop and Spark
Developing and Deploying Analytics for IoT Systems
The combination of smart connected devices with data analytics and machine learning is enabling a wide range of applications, from home-grown traffic monitors to sophisticated predictive maintenance systems and futuristic consumer products. While the potential of the Internet of Things (IoT) is virtually limitless, designing IoT systems can seem daunting, requiring a complex web infrastructure and multinomial expertise.
In this session, Amit discusses how to prototype and deploy an IoT system with data analytics without developing custom web software or servers. The workflow is based on MATLAB® and ThingSpeak™, an analytic IoT platform that can run MATLAB code on demand in the cloud.
- Conceptual overview of ThingSpeak
- Collecting and storing sensor data
- Integrating online analysis and visualization using MATLAB
- Scaling IoT solutions with analytics
Parallel Computing with MATLAB and Simulink
Large-scale simulations and data processing tasks take an unreasonably long time to complete or require a lot of computer memory. Users can expedite these tasks by taking advantage of high-performance computing resources, such as multicore computers, GPUs, computer clusters, and cloud computing services.
In this session, Alka discusses how to boost the execution speed of computationally and data-intensive problems using MATLAB® and parallel computing products. Alka demonstrates several high-level programming constructs that allow you to easily create parallel MATLAB applications without low-level programming.
- Learn high-level programming constructs as well as built-in parallel algorithms to solve computationally and data-intensive problems using multicore processors and GPUs
- Scale up to clusters, grids, and clouds using MATLAB Distributed Computing Server™ with minimum programming efforts
- Run multiple simulations of a model in parallel
Simulink as Your Enterprise Simulation Platform
Engineers are increasing their adoption of system simulation to develop the complex integrated systems needed in today’s market. Gone are the days when embedded software can be written and tested directly on physical prototypes. An environment that can model both the algorithmic and physical components of a system is needed to fully understand and develop the systems of tomorrow.
Simulink® is an enterprise simulation platform that meets those needs. With its scalable multidomain modeling and simulation capabilities, you can author components using both textual and graphical elements—including MATLAB® functions and objects, block diagrams, state machines, and flow charts—and simulate discrete, continuous, discrete-event, and physical systems. An enterprise simulation platform also needs to scale to make teams more efficient in working together and in simulating large systems that consume and produce massive amounts of data. Finally, an enterprise simulation platform also needs to be able to integrate third-party IP to address specific component modeling needs.
Join this session to discover how to use Simulink as an enterprise simulation platform.
Development and Deployment of ECU-Based Control Systems Using Model-Based Design
Often, tackling a problem faced by customers calls for innovative and outside-the-box solutions, implemented with minimal time-to-market. This is increasingly challenging for lean organizations employing small teams. The standard V-model process can be adapted to suit specific needs. However, the scale and scope of development should be kept simple enough to be maintained and serviced by small teams.
This presentation explores the suitability and various usage aspects of model-based software development, and how small teams can benefit from it. Model-Based Design is highly friendly to development of complex systems, which are usually derived from simple, non-technical customer requirements. Implementation of such solutions requires that there is good traceability between requirements and final solution and that design best practices, such as maintainability and scalability of the solution over time, are held intact.
Suchin discusses how TAFE utilizes Simulink® and Model-Based Design to develop an electronic control system, covering various phases such as requirements tracing, modeling, and code generation in the process.
Generating Optimized Code for Embedded Microcontroller Algorithms
Embedded code generation is fundamentally changing the way engineers work. Instead of writing thousands of lines of code by hand, engineers are automatically generating their production code to increase productivity, improve quality, and foster innovation. Because code generation is a proven approach for dealing with the ever-increasing complexity of embedded software algorithms, the need to automatically generate code is increasing daily. One challenge every embedded code generation or C coding engineer faces is reducing the effort required for implementing optimized code into their resource-constrained, mass-production microcontrollers.
Join this session to learn about:
- The latest features in Simulink®, Stateflow®, and Embedded Coder® for generating highly optimized code
- Automating the process for generating algorithm code that plugs into and integrates with device drivers, schedulers, and build processes
- Tuning and monitoring controller parameters and variables in real time
Modeling Mechanical and Hydraulic Systems in Simscape
Do you still rely on hardware prototypes for designing your mechanical or hydraulic systems? Do you also face the challenge of unifying multiple domains to simulate the performance of the entire system? Join this session to learn how Simscape™ helps engineers “reach for the run button” and enables them to use simulation to save time and money. You will also learn how Simscape and its add-on libraries enable engineers to model and simulate a wide range of systems, including multibody systems and fluid power systems.
You will learn how to:
- Easily build Simscape models of the physical systems
- Import CAD models for reusing it in Simscape platform
- Perform simulation modes for analyzing motion and calculating forces
- Model hydraulic systems with components such as valves, cylinders, and pipelines
- Leverage MATLAB® capabilities for finding optimal designs
Verification, Validation, and Test in Model-Based Design
Applying verification and validation techniques throughout the development process enables you to find design errors before they can derail your project. Most system design errors are introduced in the original specification but are not found until the test phase. When engineering teams use models to perform virtual testing early in a project, they eliminate problems and reduce development time.
In this session, Manohar discusses how you can apply early verification and validation activities at every stage of the development process in Model-Based Design.
- Detecting design errors in models using formal verification methods
- Systematic simulation testing of design by using an automation framework
- Using model slicing to analyze and debug problematic behavior in a model
- Automatically generating reusable tests that satisfy model and code coverage
Technologies for Developing Smart Systems
Designing and Implementing Real-Time Signal Processing Systems
Signal processing is essential for a wide range of applications, from data science to real-time embedded systems. Some of the challenges in developing signal processing system are the acquiring and processing raw data from sensors to derive meaningful information and designing algorithms for real-time processing. MATLAB® and Simulink® provide a platform for exploring and analyzing time-series data and a unified workflow for the development of embedded DSP software and hardware by providing a complete workflow for fixed-point design and C and HDL code generation.
During this session, you will learn about:
- Acquiring, measuring, and analyzing signals from various sources
- Designing streaming algorithms to analyze patterns in signals and extract meaningful information
- Implementing, prototyping, and testing DSP algorithms on embedded processors, SoCs, and FPGAs
Spectral Imaging: Breast Density Measurement Using MATLAB Coder
Abha’s division at Philips primarily works on imaging and research, focusing on spectral breast density measurement.
Her presentation discusses developing a reliable and accurate breast density measurement (BDM) algorithm in MATLAB®, moving it into a production environment by porting the MATLAB code to C++ quickly, and addressing the continuous changes in MATLAB code.
She also shares her results, which include faster code conversion in MATLAB to C++ that produces accurate results with MATLAB output.
Developing Autonomous Systems with MATLAB and Simulink
Image and vision, radar, EOIR, IMU, and a combination of sensor technologies are all used to automate aspects of autonomous systems. Critical functions such as object and collision detection, path and motion planning, spatial localization and mapping are designed using advanced concepts including sensor fusion and machine learning to drive guidance, navigation, and control (GNC) algorithms.
To develop these complex multidomain autonomous systems, engineers must analyze the behavior of mechanical and electrical subsystems, sensor and perception algorithms, and controls as an integrated platform, and deploy to the actual hardware. In this session, you will learn how using Model-Based Design with MATLAB® and Simulink® can help you address these challenges.
Using a quadrotor, Vivek shows how to:
- Model environmental effects and 6DOF aircraft simulations
- Develop and implement flight controls algorithms
- Design and test vision, radar, and IR perception algorithms
- Perform sensor fusion and controls development
- Connect MATLAB and Simulink to ROS environment
Developing and Prototyping Next-Generation Communications Systems
Wireless communication has seen a proliferation of standards addressing many traditional applications, such as mobile telephone and wireless broadband internet access, and emerging areas, such as Internet of Things and vehicle-to-vehicle communication. Developing radios for next-generation communications systems requires expertise in antenna and RF design, DSP and digital logic implementation, embedded software development, and system architecture modeling and simulation. MATLAB® and Simulink® provide a platform that encompasses algorithm design, system simulation, over-the-air testing, prototyping, and implementation.
In this session, Amod discusses:
- Modeling and simulating LTE and WLAN standards-compliant PHY
- Developing 5G-candidate technologies such as new modulation schemes and massive MIMO
- Multilayer modelling and simulation for MAC-PHY codesign
- Over-the-air testing and prototyping on software-defined radio platforms
Simplifying Image Processing and Computer Vision Application Development
Image processing and computer vision is an enabling technology that is driving the development of several of the smart systems today including self-driving cars, augmented reality, hyperspectral imaging, and medical imaging. Developers of modern image processing and computer vision applications face many challenges regarding handling large data sets and working with new computing paradigms, such GPU computing. You can use MATLAB® to simplify your image processing and computer vision application development workflow.
Join this session to gain insight into:
- Object detection and recognition using machine learning and deep learning
- Image processing on 3D data sets, including pixel operations, local filtering, and morphology
Automated Driving: Design and Verify Perception Systems
Automated driving systems perceive the environment using sensors, including vision, radar, and lidar, and dynamically control driving tasks such as steering, braking, and acceleration. These automated driving systems range from advanced driver assistance systems (ADAS) to full autonomy. Join this session to learn how MATLAB® can help you:
- Automate ground truth labelling tasks for deep learning
- Design sensor fusion and tracking algorithms based on logged sensor data
- Verify algorithms by synthesizing sensor data and generating traffic scenarios
MATLAB and LTE System Toolbox for Developing the LTE Physical Layer
In this presentation, T Pushpalata discusses the development and prototyping of the LTE physical layer for concept proving and capturing and adhering to system requirements. The approach included three stages of development of LTE physical layer using LTE System Toolbox™:
- Stage 1: Development of physical layer using high-level functions
- Stage 2: Development of physical layer using mid-level functions
- Stage 3: Development of physical layer using low-level functions
Project achievements included:
- Cell search procedure completed successfully.
- Broadcast message decoded in downlink.
- Control information and data decoded.
Leveraging Formal Method Based Software Verification to Prove Code Quality and Achieve MISRA Compliance
In this session, Prashant presents the use of Polyspace® products to verify critical embedded software. Polyspace products use formal methods–based static analysis to find run-time errors and prove that the software is safe. They provide comprehensive software verification capability for early stage development use, spanning bug finding, coding rules checking, and proof of the absence of run-time errors.
Polyspace Code Prover™ proves the absence of overflow, divide-by-zero, out-of-bounds array access, and other critical run-time errors in the source code. Using a unique formal methods approach called abstract interpretation, Polyspace Code Prover finds critical errors that other verification techniques can miss. Using Polyspace Bug Finder™, you can identify coding rule violations (MISRA®), programming errors, data flow problems, and other defects enabling you to triage and fix bugs early in the development process.
Through demonstrations and examples, Prashant shows how Polyspace products help detect critical run-time errors and prove that your software does not contain those errors. You will also learn how to use these verification results to certify your code to standards such as DO-178, ISO 26262, IEC 61508, and derivatives.
From Simulink to AUTOSAR: Enabling AUTOSAR Code Generation with Model-Based Design
Model-Based Design affords many advantages over traditional development by offering high-level design abstractions and automatic generation of production code. Modeling and code generation for AUTOSAR software components lets you automate the process of specifying and synchronizing lengthy identifiers in designs, code, and description files.
This session is intended for systems and software engineers who wish to understand the basic concepts, best approaches, and advanced features for using Simulink® for AUTOSAR design and Embedded Coder® for software implementation.
Durvesh will provide a brief overview of AUTOSAR standards and provide product demonstrations showing how you can use Simulink products to design, simulate, verify, and generate code for AUTOSAR application software components.
Session highlights include:
- Simulink approach to AUTOSAR
- Modeling styles
- AUTOSAR design workflows using third-party authoring tools while showing advanced AUTOSAR features
Modeling and Simulating Large Phased Array Systems
Phased arrays are widely used in today’s radar and wireless communications systems. Demand for system performance, along with advances in material and RF component technologies have made the adoption of large size arrays possible. This presentation includes a workflow for building MATLAB® and Simulink® models of large arrays of antennas as well as RF channels for both transmitters and receivers in complex electronically steered phased arrays.
In this session, Tabrez discusses how to:
- Explore alternative array and subarray system architectures and make related tradeoffs
- Synthesize array and beam patterns, including matching desired beam patterns
- Calibrate arrays with perturbations and model mutual coupling between elements
- Create RF budgets and perform analysis and simulation of large arrays
- Explore ways to partition beamforming between the digital and RF domains
- Integrate the array with spatial signal processing algorithms and channel models into a full system-level model