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 applications like predictive engine maintenance, automated trading, and medical image interpretation. 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, Mischa Kim and Stéphane Marouani show 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.
Attend this session to learn how MATLAB® can take you beyond Microsoft® Excel®. Automate your analysis workflows with thousands of prebuilt mathematical and advanced analysis functions and versatile visualisation tools. Through product demonstrations, see how to:
- Access data from files and Excel spreadsheets
- Visualise data and customise figures
- Perform statistical analysis, machine learning, optimisation, and predictive modelling
- Generate reports and automate workflows
- Share analysis tools as standalone applications, web sites, or Excel add-ins
This session is intended for people who are new to MATLAB.
Learn about the latest capabilities in MATLAB® and Simulink®. Topics include what’s new in the areas of big data, machine learning, and speeding up simulation performance. New data types in MATLAB help you work with and manage time-stamped tabular data, text data, and data that is too big to fit in memory. You’ll also see the latest enhancements that have been made to the Live Editor, where you can see results together with the code that produced them. And in Simulink, discover how just-in-time acceleration speeds up the time it takes to run your simulations in Accelerator mode.
Infrared spectroscopy exploits chemical composition and molecular and lattice structures within mineral samples to produce a characteristic response. Infrared devices, such as the TerraSpec line from ASD Inc. and the HyLoggerTM system, collect spectra in the near-infrared and short-wave infrared region from mining samples like core and rock chips. Fourier transform infrared (FTIR) instruments using a diffuse reflectance accessory collect spectra for pulp or powdered samples in the mid- and thermal-infrared range. The combination of these devices allows for the detection of hydrated, carbonate, silicate, and many oxide minerals from the spectral data.
The key to successful use of infrared spectra, however, is the interpretation methodology. Traditionally, specific spectral features are interpreted and compared with the spectra of pure minerals. Alternatively, a full pattern machine learning technique using artificial neural networks and a specific calibration set of samples can produce a model that can predict results for unknown samples. The technique significantly reduces the cost and timelines for analysis on many mining projects.
The Transformative Force of Robotics in Industry
Robotics and artificial intelligence (AI) are the next transformative technologies that will impact virtually every industry, from automotive to medical devices, consumer electronics to industrial manufacturing. This talk explores the current state of robotics, and discusses the various segments of industry and society where robotics technologies are expected to have the largest impact, both in the short term and not-too-distant future.
External beam radiotherapy, in which radiation dose is externally delivered to the cancerous target, is an established prostate cancer treatment modality with several unresolved issues. The goal is to utilize computational statistics to associate the 3D distribution of dose in and around the prostate target with treatment failure. This will enable the location of underdosed regions and bring about an enhanced understanding of the nature of microscopic disease spread, potentially improving patient treatment outcome.
Statistics and Machine Learning Toolbox™ and Parallel Computing Toolbox™ have enabled most of the analysis. A range of statistical and matrix manipulation functions were employed to compare the 3D dose distributions of subjects with and without treatment failure, exposing regions where dose variance is associated with failure. Furthermore, the MATLAB® Classification Learner app has been extremely useful in the preliminary establishment of a model based on machine learning that can classify participants in terms of treatment failure (and potentially side effects). MATLAB has also assisted in the visualisation of 3D dose distributions and all 3D results.
Automatic 3D Human Action Recognition
Automatic recognition of human actions in videos is useful for smart surveillance, security, elderly care, child minding, and human machine interaction. However, capturing labelled training data for this task can be prohibitively time-consuming and expensive. To overcome this problem, we synthesize 3D human poses and render them from a large number of camera viewpoints in MATLAB®. We then use the large corpus of training data to train a deep convolutional neural network (DCNN) in MATLAB that learns a camera viewpoint independent representation of the human pose. By modelling the temporal changes in human poses using Fourier Temporal Pyramids, we can achieve state-of-the-art results for 3D human action recognition.
This research is part of a Discovery grant on smart surveillance funded by Australian Research Council.
A Machine Learning–Based Speech Processing Solution for Facilitating Early Diagnosis of Parkinson's Disease
Recent advances in speech processing have facilitated the assessment of prognosis of patients with Parkinson’s Disease. However, there is currently no objective method to provide an early diagnosis of Parkinson’s Disease. This presentation discusses how the Auckland Bioengineering Institute at The University of Auckland used MATLAB® for fitting and pattern recognition to determine a minimum viable solution for aiding early diagnosis of Parkinson’s Disease.
Detecting Moving Objects in Aerial Imagery Captured from Unmanned Aerial Vehicles
Detecting moving objects in video footage is a fundamental preprocessing step involved in object detection and tracking. It is used in many real-time applications such as surveillance and traffic monitoring. Detecting moving objects in videos captured from a camera mounted on an unmanned aerial vehicle (UAV) is a challenging task. The camera is constantly moving in addition to the moving objects, which is why there is a need to distinguish the movement of the camera, global movement, from the movement of the foreground, local movement. UAV data suffer scale variations (due to altitude change of the UAV) and viewpoint variations (due to roll, pitch, and yaw), and this needs to be addressed when compensating for global movement.
As part of the research in object detection and tracking, a camera movement compensation algorithm has been implemented successfully using MATLAB®, Image Processing Toolbox™, and Computer Vision Toolbox™. The method involves estimating the camera movement between frames using the Kanade-Lucas-Tomasi (KLT) point tracker algorithm and correcting for it. Once the camera movement is corrected, frame differencing and background modelling methods are used to detect moving foregrounds. Beneficial results were achieved with minimal coding and faster turnaround time by using the built-in methods available in MATLAB.
RoboBEER: Using MATLAB and Computer Vision to Analyse Beer Quality
Visual characteristics of beers, such as foam, colour, and bubbles, are the first considerations that consumers evaluate when judging quality. Several methods have been developed to assess foaming parameters, which are usually not standardised, and can be time-consuming and costly. For this reason, a robotic pourer prototype named RoboBEER was developed using LEGO® blocks, servo motors, open hardware sensors, and Arduino® boards.
This presentation discusses how MATLAB® and low-cost hardware were used to classify beer quality based on foamability and colour analysis.
Using MATLAB and Machine Learning to Develop Respiratory Disease Diagnostic Tools
ResApp Health is developing instant and accurate respiratory disease diagnostic tests and management tools for smartphones. Dr. Udantha Abeyratne at The University of Queensland originally developed ResApp’s technology. The technology is based on the premise that cough sounds carry vital information on the state of the respiratory track. Using signatures in cough sounds, the technology can accurately diagnose a wide range of chronic and acute respiratory diseases such as pneumonia, asthma, and bronchiolitis.
Using MATLAB®, the team at The University of Queensland and ResApp’s engineers developed end-to-end machine learning algorithms to detect and analyse cough sounds. The first part of the system finds cough sounds in an audio stream, ignoring other sounds such as speech and background noise. After detection of the cough sounds, each cough is analysed for a signature associated with a respiratory disease. ResApp’s engineers developed the system using Signal Processing Toolbox™ and Statistics and Machine Learning Toolbox™. The final algorithms are then deployed on various smartphone platforms and operate in real time.
Achieving Measurable Business Results Partnering with MathWorks: Practical Examples from Around the World
11:45 a.m.–12:15 p.m.
This presentation focuses on MathWorks Consulting Services. Kevin Rzemien and Branko Dijkstra share customer success stories and how companies saved time and money while learning more about MATLAB® and Simulink®. Practical examples will be presented, including the challenges, the implemented solutions, and the overall results of the consulting engagements.
Data is everywhere, and each year users store more and more of it. Huge data sets present an amazing opportunity to discover new things about the world, the products made, and how people interact with them. However, big data sets also present some real challenges. How do you understand them? How do you interrogate them? How do you even read them? In this talk, David discusses new tools in 2016 MATLAB® and Simulink® product releases that help you work with even bigger data.
Analytics has become central to setting business strategy, enabling tactics, and driving operations. These are different challenges and time scales, and require visual and predictive analytics to support decisions and automate actions. Spotfire’s visual analytics and the numerical algorithms of MATLAB® are the best in class, and the two software systems combine very naturally to form a productive platform for analytics app development.
This presentation showcases the Spotfire-MATLAB analytics platform, and describes case studies across a broad set of industries. Examples from both sides of the brain—from engineering and marketing—are discussed. The focus is on enterprise apps that are deployed at scale and driving extreme business value. Highlighted examples include context from the finance, energy, transportation, and manufacturing industries.
Deep learning is ubiquitous. From medical diagnosis, speech, and object recognition to engine health monitoring and predictive maintenance, deep learning techniques are being used to make critical engineering and business decisions every moment of the day. In this session, David discusses deep learning techniques in MATLAB®, and in particular, addresses the computer vision problem of object detection and recognition.
Predictive maintenance—the practice of forecasting equipment failures before they occur—is a high priority for many organisations looking to get business value from historical performance data. New technologies, such as machine learning and big data, show promising results, but they fail to capture nuances that may be obvious to domain experts familiar with these systems.
In this session, you will see how you can use machine learning and big data techniques with traditional techniques for Model-Based Design to create hybrid approaches for predicting failures. Through examples and case studies, Daryl shows how MATLAB® and Simulink® combine to provide a common platform for building predictive maintenance algorithms.
Engineers are increasingly adopting system simulation to develop complex integrated systems. Gone are the days when embedded software can be written and tested directly on the physical prototypes. An environment that can model both the algorithmic and physical components of a system helps engineers fully understand and develop the systems of tomorrow.
Simulink® is an enterprise simulation platform with scalable multidomain modelling and simulation capabilities. With Simulink, you can author components using both textual and graphical elements, and simulate discrete, continuous, discrete-event, and physical systems.
Simulink scales to make teams more efficient in working together and in simulating large systems that consume and produce massive amounts of data. This presentation highlights how you can use Simulink to address the challenges of simulation performance, model complexity, big data, and team development.
Finally, Simulink integrates third-party IP to help you address specific component modelling needs. You’ll see how Simulink can cosimulate with over 100 third-party partners in the MathWorks Connections Program.
Join this session to discover how you can use Simulink as your enterprise simulation platform.
Computational thinking is a problem-solving process that helps students learn new things by making connections between the concepts they already understand and the application of that knowledge. When done right, a computational thinking approach not only shows a student how to get from A to B, but it can also give them the confidence and desire for continued learning and problem solving.
In this session, you’ll see how the MATLAB® technical computing environment supports computational thinking. Specifically, you’ll learn how to:
- Decompose big problems into a sequence of smaller ones
- Implement a variety of programmatic thinking techniques, including symbolic representation and numeric solution
- Compose a learning narration that includes problem formulation, solution, and analysis.
Developing an Internet of Things (IoT) system involves embedded programming and cloud-based analytics. Get started quickly by using MATLAB® and Simulink® to deploy algorithms on your smart embedded devices, collect and analyse data in the cloud, and build predictive algorithms.