Competitions prepare students for changing technology landscapes

August 10, 2017 Sebastian Castro, MathWorks

The days when engineering teams worked in silos, each contributing individual skills, are rapidly evolving into a new working environment where success is determined by breaking these silos down to support tool integration. Newer concepts such as deep learning, cyber-physical systems, and the IoT are demanding that engineers work across multiple disciplines to produce unmanned aerial vehicles (UAVs), self-driving cars, or other products to help fill bigger market needs. For example, McKinsey & Company predicted that up to 15 percent of new vehicles sold by 2030 could be fully autonomous.

Many of these technologies, such as "smart" sensors, vision-based systems, and machine learning algorithms, are still in the research phase. A Forrester study found that 58 percent of enterprises are researching artificial intelligence, but only 12 percent are using it in products.

Higher education has taken note. Typically a bastion of research, academia is turning to student competitions such as RoboCup to offer students the opportunity to continue to explore cutting-edge technologies outside of the traditional classroom, in a more real-world setting.

RoboCup incorporates robotics research into a long-term mission to build a team of autonomous soccer-playing robots that are capable of beating a squad of humans by 2050 while also helping students understand robotic systems. Founded in Japan, the decades-long competition has spawned sister competitions, such as RoboCupRescue, whose technology has been adopted by first responders for search-and-rescue missions, and RoboCupIndustrial, whose technology has been adapted to refine manufacturing services and factory logistics.

What’s unique about these competitions is that engineering students are asked to design, prototype, and build autonomous vehicles and robots using standard industry hardware and software tools, including MATLAB and Simulink, in much the same way they would if they were working for a commercial enterprise.

[Figures | (1) Simulation of a ground robot performing autonomous sign detection. (2) Simulink model containing computer vision, machine learning, and control algorithms. The model uses the Robotics System Toolbox to communicate with the simulator. (3) Image from simulated robot camera with detected sign labels, “speed limit” and “membrane.”]

Within RoboCup, students gain knowledge of multiple technologies, including:

  • Computer vision and machine learning, where the focus is on autonomously detecting, classifying, and tracking objects of interest
  • Motion planning and controls, where the focus is on programming robots to stably and reliably move around their environment, while efficiently using and storing energy
  • Cyber-physical systems, which may include multirobot collaboration, communication between robots and humans, and leveraging cloud computing and IoT services for data-driven tasks like speech recognition

All of these technologies are challenging on their own, but the combination of them is very difficult yet necessary for total autonomy.

While RoboCup is an example of the gamification of engineering, the competition’s results are at the forefront of robotics research technology. Each year, teams competing from around the world develop autonomous robots with the potential to address real-world problems. For example, RoboCupRescue teams are challenged to design wheeled robots that can maneuver in post-disaster scenarios with rough terrain. The various leagues in the RoboCup competition give students the opportunity to work with real robots and experience challenges that are common in the classroom, like working on a deadline and collaborating with other students.

Potential employers are paying attention

Building competitions around mastering concepts like computer vision, machine learning, and cyber-physical systems yields technologies that can be immediately applied to commercial robotics applications while molding student competitors into a better prepared, more versatile workforce. And while the early development of practical design experience is changing the way students are taught, it’s also catching the attention of prospective employers who are looking for the next generation of hybrid engineers.

MathWorks, along with hundreds of other technology companies, is spending more time and resources investing in competitions as a path to accelerated learning. Sponsoring companies build relationships with participating teams as guides and mentors. MathWorks, for example, supplies students with licenses for MATLAB and related toolboxes, assigns company representatives to conduct training sessions, and provides on-call assistance to answer questions and provide feedback.

These concepts aren’t new, but the ability to combine control systems and image processing to create an autonomous vehicle and robot requires a new approach to cross-functional engineering. RoboCup’s competitions are building the next generation of versatile engineers.

Sebastian Castro is an Education Technical Evangelist at MathWorks, supporting university-level robotics competitions such as RoboCup. In previous roles, he created training material for MathWorks modeling and simulation solutions, and worked on system-level simulation and design of a conceptual solar-powered UAV. Sebastian has a MS degree in Mechanical Engineering from Cornell University, where he researched high-level control of modular robots.

Previous Article
PWM control with the Raspberry Pi

Pulse-Width Modulation (PWM) is a modulation technique that is widely used to control power. It is commonly...

Next Article
Making machines move when an MCU isn't the answer

If you're developing a device that needs to make something move at a set interval, the obvious solution for...