Workshops

Below are the details of the workshops/tutorials that have currently been approved.

Computer Vision using Machine Learning with MATLAB (morning)

Time: 8h00-12h00

Venue: Civil Engineering Building, S203

Organizers/presenters: Kirsten Smith, Michelle Navarro, and Praneet Kala (Opti-Num solutions)

Objectives and content: Computer vision is an enabling technology that is driving the development of several of the smart systems today—for example, self-driving cars, gesture-based gaming, and robotic surgery systems. Computer vision applies complex algorithms to images and video to detect, classify, and track objects or events in order to understand a real-world scene. In this session, Opti-Num Solutions engineers will show you how MATLAB can be used to simplify the computer vision application development workflow from idea to deployment.Through demonstrations, you will gain insight into:

  • Leveraging MATLAB for computer vision application development
  • Image Classification
  • Tracking Objects of Interest
  • Using Stereo Vision
  • Generating C-code from MATLAB

Demo detail: Robotics System Toolbox provides an interface between MATLAB and Simulink and the Robot Operating System (ROS). This enables you to communicate with a ROS network, interactively explore robot capabilities, and visualize sensor data. ROS is a communication interface that enables different parts of a robotic system to discover, send, and receive data.  MATLAB support for ROS is a library of functions that allows you to exchange data with ROS-enabled physical robots, or robot simulators such as Gazebo.

The demo that will be showcased explores MATLAB control of the Gazebo Simulator using ROS. When using robot simulators, it is important to test autonomous algorithms and dynamically alter the surroundings in the world model while the simulation is running. The demo will show how to create basic robot autonomy with Gazebo and how to interact with it. In this example the robot is the TurtleBot platform. A timer class will be used to control the autonomous aspects of TurtleBot movement. While you can use loops and other methods to examine basic autonomy, the scheduled execution and non-blocking nature of timers make them the best choice for achieving autonomous behaviour.

Intended audience: Research Areas and Industries: Aerospace and Defence, Automotive, Electronics, Medical Devices, Education, Computer Science, Earth and Ocean Sciences, Technical Services and Consulting.


Scalable Data Science and Distributed Computing with MATLAB (afternoon)

Time: 13h00-17h00

Venue: Civil Engineering Building, S203

Organizers/presenters: Kirsten Smith, Michelle Navarro, and Praneet Kala (Opti-Num solutions)

Objectives and content: MATLAB is a scalable platform for supporting Data Analytics – a rapid prototyping environment which enables you to access and analyse data and develop predictive models. MATLAB works with your existing systems, so you can optimally process your algorithms – without having to rewrite them – using anything from your local desktop machine to Hadoop and Spark, to improve the execution efficiency of your code. This workshop will highlight combining MATLAB’s MapReduce framework with its Machine Learning capabilities. In this session, Opti-Num Solutions engineers will show you how MATLAB can be used for scaled performance and working with big datasets, even if they don’t fit in memory.Through demonstrations, you will gain insight into:

  • Leveraging MATLAB for scaled Data Science applications
  • Parallel Computing
  • Statistics and Machine Learning Toolbox

Demo detail: Robotics System Toolbox provides an interface between MATLAB and Simulink and the Robot Operating System (ROS). This enables you to communicate with a ROS network, interactively explore robot capabilities, and visualize sensor data. ROS is a communication interface that enables different parts of a robotic system to discover, send, and receive data.  MATLAB support for ROS is a library of functions that allows you to exchange data with ROS-enabled physical robots, or robot simulators such as Gazebo.

The demo that will be showcased explores MATLAB control of the Gazebo Simulator using ROS. When using robot simulators, it is important to test autonomous algorithms and dynamically alter the surroundings in the world model while the simulation is running. The demo will show how to create basic robot autonomy with Gazebo and how to interact with it. In this example the robot is the TurtleBot platform. A timer class will be used to control the autonomous aspects of TurtleBot movement. While you can use loops and other methods to examine basic autonomy, the scheduled execution and non-blocking nature of timers make them the best choice for achieving autonomous behaviour.

Intended audience: Research Areas and Industries: Aerospace and Defence, Automotive, Electronics, Medical Devices, Education, Computer Science, Earth and Ocean Sciences, Technical Services and Consulting.


HLT in South Africa: State of the community (afternoon)

 

Please see the invitation to participate in/contribute to this workshop.

Time: 13h00-17h00

Venue: Electrical and Electronic Engineering Building, E352

Organizers/presenters:

Organizers: Dr. Roald Eiselen (CTexT, NWU; Resource Management Agency); Dr. Febe de Wet (Meraka agency, CSIR); Dr Daniel van Niekerk (MuST, NWU).

Invited speaker: Prof JC Roux (South African Centre for Digital Language Resources)
Other confirmed presentation/demonstration groups:

  • HLT Research Group, CSIR Meraka Institute.
  • Centre for Text Technology, North-West University.
  • Multilingual Speech Technologies, North-West University.

Objectives and content: Human language technology (HLT) remains an important enabling technology, which allows greater access to information in the digital age. This is especially important in the South African context with eleven official languages and relatively low levels of access to information in people’s first language. The development of HLTs and resources for the South African languages has come a long way over the last fifteen years, with major advances in the quantity and quality of the resources that are available for research and development. These advances have been supported by several large scale development projects in both speech and text processing, including the African Speech Technology, Lwazi, NCHLT Speech, NCHLT Text and Autshumato projects. There have also been several other projects and individual efforts to extend and improve the technologies and resources that are available for the automatic processing of South African languages.

Even with these efforts, most of the official South African languages are still considered under-resourced in the scope of HLT and more work is required to ensure the continued development and improvement of HLT in South Africa. The matter is further complicated by the fact that that HLT is a cross-disciplinary field that requires knowledge of at least linguistics, computer science, and statistics. Because the community of HLT researchers and developers is relatively small, it is important that the various research groups coordinate their development efforts to complement each other to ensure that the limited resources available for development are effectively utilised.

This workshop aims to bring together South African researchers and developers of HLT applications and resources to present, demonstrate and discuss both their previous and current projects, as well as projecting what they foresee their focus will be for the next three to five years. The workshop has an invited presentation by a representative of the newly established South African Centre for Digital Language Resources, a national infrastructure for the development and distribution of resources for the South African languages, as well as confirmed presentations and demonstrations by three HLT groups, with an open invitation to all other research institutions and individuals to present and demonstrate their HLT technologies.

The workshop will also include an open forum for the discussion of possible cooperative and coordinated development efforts with topics including, but not limited to:

  • “grand challenges” for HLT in the South African context;
  • challenges of developing HLT for resource-scarce languages;
  • data, tools and applications for HLT development;
  • language technology integration;
  • evaluation of HLT technologies and resources; and
  • funding opportunities for the development of language resources and technologies.

Intended audience: The intended audience includes researchers and developers from various institutions that currently work on HLT in the South African context. PRASA provides the ideal environment for getting together the larger HLT community from around the country and discuss the issues they encounter, as well as creating an opportunity for dialogue on national challenges and planning the path toward greater cooperation between the different groups across the country. The workshop will also provide individual researchers the opportunity to discuss and align their research and development with some of the larger groups with the aim of extending the current HLT base to a larger number of researchers and institutions.


Functional Safety: The Pathway to Safe Automated Machines using Siemens TIA Portal (full day)

Time: 8h30-15h00

Venue: Electrical and Electronic Engineering Building, E202

Organizers/presenters: Prof. Theo van Niekerk and Mr. John Fernandes (Advanced Mechatronic Technology Centre, NMMU)

Objectives and content: The objective of the workshop is to introduce delegates to the philosophy of functional safety and integrated safety solutions for the development of safe automated machines.  The workshop will cover various aspects of Functional Safety within the machine design process including:

  1. The importance of ‘safe machinery’
  2. Relevant European safety standards for machine builders
  3. The pathway to conformity (CE marking and certification)
  4. Procedures used in determining and eliminating risk according to EN ISO 12100
  5. Calculating various safety parameters such as the required Performance Level (PL) and Safety Integrity Level (SIL)
  6. Methods used for verifying and comparing the achieved safety levels according to IEC 61508 and IEC 62061.
  7. The design of suitable safety circuits according to EN 954-1 and its successor EN ISO 13849-1.
  8. The role of Programmable Logic Controllers (PLCs) in achieving acceptable safety levels within machines and processes as well as the importance of ‘safe’ communication between systems.
  9. Practical session on integrated safety using Siemens TIA portal
    a. Device setup and configuration
    b. Practical examples.

Intended audience: Nowadays, almost all industrial processes incorporate electrical and electronic devices and systems. Many of these could cause severe harm to workers or the environment if not for special ‘integrated’ safety features. Functional safety is thus aimed at all who design, manufacture, use, maintain, modify, manage or inspect machinery or processes. Functional safety begins by helping manufacturers to systematically identify all possible risks associated with their machines or products throughout the machines entire life-cycle and makes recommendations on how these risks can be reduced to tolerable levels through the addition of safety features such as safeguarding, presence sensing, interlocking and computer diagnostics. Functional Safety is applicable to almost all areas of industry including the oil and gas industry, the manufacturing and transportation sectors and nuclear plants just to name a few.


An introduction to causal models, Bayesian networks, and particle filters

Organizers/presenters: Dr. Alta de Waal (University of Pretoria) and Dr. Gregor Pavlin (Thales Research and Technology, The Netherlands)

PRASA-Robmech delegates are also invited to attend the final morning of the FAIR 2016 workshop, which will cover an introduction to causal models, Bayesian networks, and particle filters.  (The theme of the FAIR 2016 workshop is “CAUSALITY:  Bayesian Networks as Platform for Knowledge Representation in Science and Philosophy”.  For more details on the sessions on the final day, as well as the rest of the workshop, please see http://www.cair.za.net/FAIR2016/.  Registration is free, but must be done on the FAIR workshop website.