THE STORY IN BRIEF
  • RAIBO2 finished the November 2024 marathon in 4 hours, 19 minutes and 52 seconds on a single charge.
  • A Nature study published in September 2026 analyses the design and field performance.
  • Energy consumed in supporting and moving the legs makes endurance a whole-machine question.
MAKE TIME FOR A GOOD READ
Inside this story 2 perspectives

The marathon runner had four legs, no need for a drinks station and no opportunity to exchange its battery. In November 2024, RAIBO2 covered the 42.195 kilometres of the Sangju Dried Persimmon Marathon in South Korea in 4 hours, 19 minutes and 52 seconds. The new result this week is the engineering account of that run. A paper published in Nature on 23 September explains what it took for the robot designed at KAIST to reach the finish on one charge.

A marathon is a revealing test for a legged machine because a small loss repeated at every footfall can add up to a very large one. Motors support the body while the legs swing, land and push away again. A machine that runs beautifully for a minute can still waste enough energy to stop well short of a useful working distance. The exercise asks what happens when the same pattern continues beyond the length of a laboratory demonstration.

Measure the whole machine

A bigger battery is the obvious idea. It is also more mass for the legs to carry. A stronger motor can deliver more force, but both its weight and its operating conditions affect the energy needed elsewhere. A change in a joint can alter the way the controller times a step. The KAIST team analysed the mechanical and electrical systems together with the way the robot moves.

The results therefore address an engineering trade-off that a component catalogue alone cannot settle. Replacing one part with a more efficient version may help at its preferred operating point. It matters whether the moving robot actually spends much time at that point, how much extra weight the replacement brings and what happens to the other joints. The test on a public-road course gave the engineers hours of operation under conditions a short, level track would not reproduce.

What the finish line proves

The run establishes a particular result for one system on one course. It does not tell an operator how long a similar machine could work while carrying instruments, stopping repeatedly or climbing a different route. Speed, payload, terrain and opportunities to recover energy all affect that calculation. A site manager choosing equipment for a long inspection would need evidence from an outing closer to that job.

KAIST says a university-linked company is working on taking the research towards commercial production. Repeating a single prototype’s performance across a production run is a further engineering problem. The marathon nevertheless gives prospective users something more meaningful than a peak speed or a battery rating in isolation. It gives a measured distance, a time and the record of a machine that carried its own energy all the way.

BB
ABOUT THE AUTHOR

Beyond Business Editors

Editorial desk

Beyond Business connects global developments with the decisions, ambitions and everyday lives of its readers. Our editorial desk explores the stories shaping our world.

More from this author
Keep the perspective.