NewsMacroThe 'Robot Olympics' Were Spectacular — But Will This Technology End Up Helping Humans?

The 'Robot Olympics' Were Spectacular — But Will This Technology End Up Helping Humans?

Author: Bworldonline·

Key Takeaways

  • •The second World Humanoid Robot Games in Beijing featured over 2,000 humanoid robots competing in 50 events, a major increase from the first edition's roughly 500 robots in 25 events.
  • •Two robots beat Usain Bolt's 100-meter world record, with the fastest finishing in 9.39 seconds, and a robot completed the 400m in 39.7 seconds.
  • •Non-sporting events such as book sorting, garden management, bricklaying, and tweezer bean picking test real-world skills and expose the difficulties robots face with everyday tasks.
  • •Companies including Tesla, Figure AI, and Unitree have attracted billions of dollars in investment, and China has designated humanoid robotics as a priority industry.
  • •Practical future uses for humanoid robots include unloading trucks, stocking shelves, working in hazardous environments, and supporting elderly independence amid care worker shortages.
The 'Robot Olympics' Were Spectacular — But Will This Technology End Up Helping Humans?

By Jonathan Roberts

Over the past few days, social media feeds may have been full of videos of humanoid robots sprinting, jumping, and performing various incredible — and not so incredible — feats. The reason: the second World Humanoid Robot Games have just finished in Beijing.

The event, hosted by China as part of its broader push to lead the global robotics industry, featured many remarkable sights, but some of the more mundane competitions offered greater insight into how this technology might eventually help humans.

Robots Are Now Faster Than People

Commonly referred to as the robot Olympics, the event began last year, when roughly 500 robots demonstrated their abilities across 25 events. This year's edition featured more than 2,000 robots competing in 50 events.

Entrants must be so-called humanoid robots: under the games' rules, they need a torso, two arms, and two legs, an onboard power source, and no trailing cables.

Half of the events are direct copies of the human Olympics and Paralympics: track and field, weightlifting, soccer, and boxing.

As in the human Olympics, the crowd-pleaser was the 100-meter sprint. This year, two robots beat Usain Bolt's world record of 9.58 seconds, with the fastest clocking 9.39 seconds. The same story played out in the 400m, where one robot finished in 39.7 seconds — the current men's world record is 43.03 and the women's is 47.60.

Impressive stuff. But it is the other 25 events, which do not resemble the human Olympics and Paralympics, that paint a more interesting picture.

A New Kind of Competition

The list of non-sporting events is very telling about how humanoid robotics is developing — or how the robots' manufacturers hope it will develop. That list includes book sorting, garden management, bricklaying, and possibly the most exciting event of all: tweezer bean picking.

These are clearly skills based on real-world scenarios outside the sporting arena, and that is what makes these games so interesting. They echo the kind of benchmark competitions that have historically accelerated progress in robotics — the DARPA Robotics Challenge in the mid-2010s similarly exposed how difficult it is for machines to perform seemingly simple tasks like opening doors or climbing stairs in unstructured environments.

Getting a robot to run 100m in a straight line very quickly is a great engineering achievement. Developing a robot that can jump 2.8 meters into the air from a standing position is also impressive. But what about a humanoid that can make you a cup of tea in your own kitchen, or properly clean out the cat's litter tray? Such tasks are currently science fiction in the world of robotics. Sending people to the Moon is more straightforward than developing a humanoid robot that can reliably load and unload a dishwasher.

Most people can easily use a dishwasher, but if you break down what you are actually doing when carrying out this everyday task, it is extremely complicated. It involves remarkable sensing, sophisticated planning, and great object manipulation — from plates, to teaspoons, to perhaps even pots and pans (if you are that kind of all-in dishwasher person).

Humans do not have to think too hard about these things, but this is precisely what robots struggle with, because they have a very limited ability to make fast decisions to cope with a world that changes quickly and unpredictably. A plate may slip in a robot's hand, meaning it must adjust its grip. A cluttered dishwasher is a minefield of failure opportunities.

For a humanoid robot to reliably complete a complex task, it typically needs first to see the objects it must interact with. This alone is incredibly difficult. Lighting conditions that do not trouble humans remain difficult for robots.

If a robot sees the object it needs to pick up, it must then work out how to pick it up, which might involve judging both large and small forces. Again, this is very hard for a robot to do. It must determine where and how to move the object, perhaps put it down, and then plan the next move. The sequence of actions is very challenging to get right every time, given that each step is difficult in its own right.

And the hardest part? A robot must recognize when something has not gone according to plan and then work out what to do next.

It is therefore the non-sporting events at the Humanoid Robot Games that are the true test of how well humanoids are progressing toward becoming useful tools. This year's games have been exposing exactly these sorts of difficulties.

Some of the most revealing challenges are the small imperfections of the real world: a cable at the wrong angle, an object slightly out of reach, or a package that shifts unexpectedly. Robots need to recognize these problems and recover without a human stepping in.

The Future for Humans — and Robots

It is easy to see why these competitions are useful to technology companies. A spectacular sports demonstration can show what a robot can do when everything is easy and goes right, while real-world competitions based on everyday tasks can expose what happens when things go wrong.

The stakes are considerable. Companies including Tesla, which is developing its Optimus robot, and startups such as Figure AI and Unitree, one of the most prominent Chinese entrants in the field, have attracted billions of dollars in investment on the premise that general-purpose humanoid robots will one day work alongside people. Governments are betting on the same outcome, with China identifying humanoid robotics as a priority industry.

These types of competitions also reveal an important truth about the current humanoid robot boom: there is still a very long way to go before we might see a humanoid robot in our homes or assisting us with day-to-day tasks. But the vast sums of money now being invested mean that day may be closer than we think.

A useful future humanoid probably will not spend much time sprinting or doing martial arts. Instead, it might help unload trucks, stock supermarket shelves, work in dangerous environments, or help an older person remain independent at home — a use case that carries particular weight in countries with rapidly aging populations, where shortages of care workers are already a pressing concern.

And maybe one day, it might also do something as mundane as emptying your dishwasher.

The Conversation via Reuters Connect

Jonathan Roberts is a professor of Robotics at Queensland University of Technology. He is director of the Australian Cobotics Centre, which receives funding from the Australian Research Council as part of the ARC Industrial Transformation Training Centre program.