Skiing Robot: From Arlberg Lessons to ABB Test Arms

A skiing robot has to learn what ski instructors teach people: stay balanced on two slippery boards, read the slope and turn on command. This post follows skiing from Hannes Schneider's Arlberg ski school to the Ski Robot Challenge held beside the 2018 PyeongChang Winter Olympics, then to the skiing robots tested in China since and the labs that now measure skis by machine. It also looks at where ABB's testing cobots could fit and what still holds skiing robots back.

The humanoid ski robot DIANA, built at Hanyang University for the 2018 Ski Robot Challenge. Photo: Park et al., Sensors 2022 (CC BY 4.0)

From ski schools to a robot slalom in Korea

Skiing was long taught from person to person. Hannes Schneider, born in 1890 in Stuben am Arlberg in Austria, developed the Arlberg technique while teaching at the Hotel Post in St Anton, appeared in a 1920 documentary film about it, and had formalized his method of instruction by 1924. He co-wrote a best-selling instruction book with the filmmaker Arnold Fanck, published in English as The Wonders of Skiing in 1931. In 1939 he moved to Cranmore Mountain in North Conway, New Hampshire, where he ran the ski school until his death in 1955.

Ski instructor Hannes Schneider during his visit to Japan in 1930. Photo: unknown photographer via Wikimedia Commons (public domain)

Schneider's technique is still used as a foundation for teaching alpine skiing, and Fanck's ski films, such as The White Ecstasy, filmed at the Arlberg in the winter of 1930 to 1931, helped make skiing widely popular. Robots came much later. A Korean team wrote in 2022 that there had previously been no way for a full-sized humanoid robot to move on a snowy mountain. The first contest for skiing robots, billed as a world first, was held in February 2018 at the Welli Hilli ski resort in Hoengseong, about an hour's drive west of PyeongChang, alongside the Winter Olympics.

A jump over a crevasse in Arnold Fanck's silent film Im Kampf mit dem Berge, released in 1921. Photo: Sepp Allgeier or Arnold Fanck via Wikimedia Commons (public domain)

Eight teams from universities, institutes and companies competed for a US$10,000 prize at the Ski Robot Challenge. The rules required robots taller than 50 cm that stood on two legs, carried their own power, used skis and poles, and could bend their knees and elbows, with sensors to find the flags. In the autonomous race, Taekwon V from Minirobot of Incheon cleared all five gates in 18 seconds, and SKIRO from the Korea Institute of Robot and Convergence made a clean run in 22 seconds. Hanyang University built DIANA, a 1.23 m humanoid that used carved turns, for the same event.

Welli Hilli Park in Hoengseong County, the resort that hosted the 2018 Ski Robot Challenge, in 2015. Photo: Piotrus / Wikimedia Commons (CC BY-SA 3.0)

Robots on snow, and skis in the lab

Four years later, before the Beijing Winter Olympics, Shanghai Jiao Tong University released video of a six-legged robot with skis and poles running tests on beginner and intermediate slopes in Shenyang, including racing, turning, route planning and obstacle avoidance around people. That same month, January 2022, Hanyang's team reported tests of DIANA's carved turns on an 80 m slope with five gates and an average inclination of 15 degrees, in temperatures between minus 16 and minus 7 °C, with deep learning used to recognise the flags.

Frames from DIANA's run down a slope past a gate, from the robot's test report. Photo: Park et al., Sensors 2022 (CC BY 4.0)

In January 2026, CGTN reported that robots skied, pulled sleds and threw snowballs at the first Jingyue Robotics Ice and Snow Fun Competition in Changchun, where humanoid and four-legged robots tested their stability, precision and endurance on snow. LimX Dynamics has shown footage of its TRON robot gliding on skis in temperatures below minus 20 °C; its TRON 1 stands up to 84.5 cm tall and was designed as a platform for researchers and developers, with an open SDK. Machines are judging skis too. In 2026 SKI Magazine and the Outside Lab at CU Denver began measuring how far a ski deflects under a known load, alongside their human testers.

Alpine ski bindings of different types mounted on skis. Photo: Chianti / Wikimedia Commons (CC BY-SA 3.0)

The closest ABB work is factory testing. At Electrolux's refrigerator plant in Curitiba, Brazil, ABB YuMi and GoFa cobots carry out gas leak detection and electrical tests, and ABB says the change raised the productivity of testing by more than 68 percent. The GoFa there replaced another brand's robot and has since run with 100 percent reliability. ABB makes no skiing robots, so a ski or binding test bench would be a new job for arms like these: a GoFa, which ABB rates at 0.02 mm repeatability, could flex a ski or cycle a binding the same way again and again, with the motion planned first in RobotStudio.

An ABB YuMi cobot carries out gas leak testing on Electrolux's refrigerator line in Curitiba, Brazil. Photo: ABB

Where skiing robots go next

On holiday, the first skiing robots most people meet will probably be companions and guides. CGTN suggested that we may soon be gliding down slopes with a robot dog or taking a guided tour led by a humanoid robot. Any robot sharing a piste with families would need what the Shanghai team tested in 2022, braking control and obstacle avoidance around people, and it would have to keep working in the cold, which is what the TRON footage below minus 20 °C set out to show. Factory cobots already work at close range with people: ABB's GoFa limits its power and force with integrated torque sensors and has a rounded design without pinch points. Resorts already lean on machines at night, as our post on ski resort automation explains.

An ABB GoFa CRB 15000 cobot fitted with test tooling at Electrolux in Curitiba. Photo: ABB

In factories and labs, the bigger opportunity is testing. Ski tests rely on human testers, and two testers on the same skis can come away with different impressions of how stiff they felt, which is the gap the CU Denver machine was built to close. The lab says its data can even help brands build better skis. Its tester scans a ski's geometry and measures bending under load, and robot arms could take that kind of rig further, from single bending tests to repeated, programmed movements that copy how a skier loads a ski.

An ABB cobot beside the gas leak detection equipment on Electrolux's refrigerator line. Photo: ABB

Most of the open problems showed up in 2018. The Welli Hilli course was steep and slippery, the light was bright enough to trouble vision systems, and the wind that postponed Alpine events at the Olympics nearby also made the slope harder. Carved turns are hard to balance, which is why DIANA's team used ZMP control to keep the biped stable, and the robot needed a waterproof design for snow. Until skiing robots cope with all of that reliably, the ones most likely to spend time on snow are those that test gear or entertain crowds.

Frames from DIANA's camera as it found the gate flags on its way down the slope. Photo: Park et al., Sensors 2022 (CC BY 4.0)

Innovation

AI solutions for effortless ABB robot control.

Automation

Robotics

ceojohntran@chatwithrobot.net

+84905311611

© 2025. All rights reserved.

qtran1215@gmail.com