By Andy Coker
Data Engineering & Data Architecture
The idea of a Robot Olympics may seem strange but with the advent of driverless cars, commercially available drones, and automated machines taking over much of the work of manufacturing and shipping, robots have never been more popular. The modern era is seeing an explosion in the number of robots and the variety of their uses. Consumers are becoming increasingly comfortable and confident around robots.
Robots are also becoming increasingly adept at tasks. As machine learning, computer vision, and other AI applications become more proficient, robots get smarter at performing specialised tasks. The cost of manufacturing robot components is also falling, making robots increasingly economically viable across many industries and use cases. As a result of these shifting perceptions and economic conditions, robotics is on the rise as a cool, interesting area of exploration for companies and average consumers alike. It’s not uncommon to hear about someone owning a drone as a hobby or building some type of autonomous system with a Raspberry Pi that accomplishes a household task.
Since robots have become more accessible, it’s worth considering whether robot competitions could be a form of entertainment for the masses. A robot Olympics or World Cup would further raise the profile of robotics and AI with the public. It would also serve as a great recruitment tool for new, young talent into robotics fields. In addition, robots are getting closer to being able to compete with human athletes, making for sports that are more high-energy and action-packed. In this article, we’ll explore the potential of a robot Olympics and what it would take for such a competition to be successful.
Humanoid Robots Competing at Human Sports
When you hear the words “Robot Olympics,” your first thought is probably human-looking robots that compete at sports that humans already play. Indeed, this is a serious area of interest for many scientists and researchers. It turns out that creating humanoid robots who are capable of complex strategy and coordination is on the cutting edge of robotics research. Getting a robot to play a sport at the same level as a human is a major challenge.
For example, researchers have held an annual robot football (soccer) competition every year since 1997. At first, the robots had difficulty standing up and even identifying the ball. Today’s competitions are not as thrilling as human soccer, but the robots have gotten much better at moving, kicking, and tracking the ball. Nevertheless, robots are still a long way from challenging human soccer players. The robots in this year’s competition still fell frequently, ran into one another, and scored few goals. The ultimate goal for this yearly competition, however, is long-term. Researchers hope to field a team of robots that can play a human team on a regulation size pitch by 2040.
Another example came from this year’s winter Olympics where the Korean host nation wanted to show off its robotics industry. Eight teams competed to send humanoid robots on a downhill skiing course, with some teams successfully completing the run and accurately weaving through all the gates. Scientists are trying robots in nearly every sport as sports tasks provide good challenges for robotics teams that could lead to future real-world applications. From badminton to basketball, robots are learning coordination, finesse, and how to work within rules and boundaries.
We’re still far from watching robots play human sports on television or in a stadium. However, progress is moving quickly and it’s not out of the question in the coming decades.
Human Piloted Competitions
Another class of competitions is events where humans and robots work together to compete. These are much closer to becoming reality than autonomous robots competing at human sports. Since humans control the robots and make decisions, researchers don’t need to program the most challenging parts of having robots compete.
Robot racing is one such sport where human pilots help with the decision making process. Far from being a simple R/C car race, these robot races involve large tracks and big cars. Some classes of competition are for class five autonomous vehicles that pilot themselves after having been trained on the course. Other races involve human pilots that must drive from afar. Imaging F1 or NASCAR but without the danger of a crash killing a human driver. As a result, drivers are far more likely to push the limits, driving faster cars through tighter turns, making for compelling competition.
The same excitement applies to the growing trend of drone racing. Pilots wear headsets and fly their drones at breakneck speeds using first-person views. The three-dimensional aspect of drone racing makes it especially exciting to watch, as pilots dodge obstacles and fly through checkpoints.
Robot Wars and Battle Bots is another type of competition where human pilots lead robots into situations where humans otherwise couldn’t go. Teams build robots designed to destroy the other robots in the competition. These shows have been on television for years, gaining a type of cult following indicating that more robot-based competitions could be popular.
With the rise of VR and other sensor technology, the types of robot sports and human inputs become more nuanced. Expect to see new types of sports and games that haven’t even been invented yet.
AI Taking on E-Sports
There doesn’t even need to be a physical game in order for technology to take on human players. The earliest AI applications worked on cerebral games like chess with IBM’s Deep Blue winning games in the 1990s. More recently, Google’s DeepMind has beaten the world’s best Go players.
Now, researchers are turning artificial intelligence to the world of e-sports to compete in video games. For example, the team at OpenAI has developed an AI that can compete at the highest levels of Dota 2, a complicated multiplayer game with a lot of strategies. Using machine learning over the course of thousands of simulations, OpenAI found new ways to challenge and beat the world’s best players.
With e-sports growing in popularity, this is a great opportunity for AI to begin to enter mainstream entertainment. E-sports is still a niche, but it’s a large and growing niche. Statista estimates e-sports viewership is up to 143 million people and predicted to reach 250 million by 2021.
Driving Innovation in Robotics & AI
The rise of e-sports and the continued popularity of shows like Battle Bots and Robot Wars seems to indicate that there’s a market for a robot Olympics. It also makes a lot of sense for universities, governments, and companies to promote these types of competitions to foster innovation and bring talent into the ecosystem. As robots continue to get better, including developing superhuman abilities, more people will want to watch the feats they perform in competition.
In fact, there are already robot competitions like a robot Olympics. The World Robot Olympiad brings together young people from countries around the world to compete in various challenges, including a football competition. Additionally, the RoboGames include over 70 events based on events at the human Olympics, including sprinting, wrestling, football, sumo, balancing, and maze solving.
Researchers feel confident that given a few decades, robots could compete with professionals in many sporting events, leading to opportunities to further bring robotics into the mainstream. While holding a robot Olympics might be profitable from an events perspective given the right media partners, its greater value lies as a research and recruitment tool. Whether the Olympiad makes money or not, investing in these competitions would drive new capabilities in robotics and encourage the younger generation to pursue robotics development.