Robohub.org
 

Solar-powered flight for 81 hours: A new endurance world record


by
31 July 2015



share this:
T=0 hours. AtlantikSolar team  hand launches AS-2 at 09:32 on July 14th, hoping to set new world records.

T=0 hours. AtlantikSolar team hand launches the AS-2 at 09:32 on July 14th, hoping to set a new world record.

Two weeks after demonstrating AtlantikSolar’s first 24-hour flight, the fixed-wing team of ETH Zurich’s Autonomous Systems Lab has reached another milestone: 81.5 hours and 2316 km of continuous flight for its 6.8kg AtlantikSolar 2 Unmanned Aerial Vehicle (AS-2). This breaks the flight endurance world record in its class.

The AS-2’s fifth test flight:

  • sets a new world record for the longest ever demonstrated continuous flight of all aircrafts below 50kg total mass, and is also the longest-ever continuous flight of a low-altitude long-endurance (LALE) aircraft (the previous record being a 48-hour flight by the 13kg SoLong UAV ).
  • is the second-longest flight ever demonstrated by an Unmanned Aerial Vehicle (behind Airbus Space’s 53kg Zephyr 7)
  • is the third-longest flight ever demonstrated by a solar airplane (behind Airbus Space’s 53kg Zephyr 7 and the 2300kg Solar Impulse 2)
  • is the fifth-longest flight ever demonstrated by any aircraft (both manned and unmanned).

In addition, the flight is a first important milestone to verify the UAV’s ability to stay airborne for multiple days while providing telecommunication services in large-scale disaster-scenarios, or live-imagery during industrial sensing and inspection missions.

T=2 hours, shortly after launch. Weather conditions: occasional clouds, and strong winds up to 40km/h.

T=2 hours, shortly after launch. Weather conditions: occasional clouds, and strong winds up to 40km/h.

Flight Summary

The flight was performed at the Rafz, Switzerland, RC-model club airfield from July 14th-17th, of which the first three days provided sunny conditions. Take-off was performed via hand-launch at 09:32 on July 14th, and after 2316km and 81.5 hours – 4 days and 3 nights – of flight, the aircraft landed safely and with fully charged batteries at 18:56 on July 17th. The fully charged batteries would in theory have enabled to continue the flight through the night again. With the exception of take off, the aircraft was in fully-autonomous operation 98% of the time, and less than 2% in autopilot-assisted mode via its Pixhawk autopilot.

T=8 hours. Monitoring the aircraft and its energy generation and storage system. The batteries are fully charged.

T=8 hours. Monitoring the aircraft and its energy generation and storage system. The batteries are fully charged.

The long-endurance flight provided helpful insights on the flight performance: The average level-flight power consumption in calm conditions (e.g. during night) was shown to be between 35-46W. Maximum power input throuh the 88 SunPower E60 cells during the day was around 260W. With this performance data, the aircraft managed to fully charged its batteries (100% SoC) at around 13:05 local time, before the time of maximum solar radiation (solar noon, occurring around 13:30). After flying through each of the three nights, the aircraft on average reached a minimum state of charge of 35% at around 07:45 local time, showing sufficient energetic safety margins for worse environmental conditions (such as longer nights, cloud cover or winds).

The flight also subjected the aircraft to a wide range of environmental conditions. Among them were thermal updrafts during the first evening/night (causing a remaining state of charge of 40% ), and downdrafts during the second night (remaining state of charge 32%). The last hours of the flight were marked by upcoming thunderstorm clouds and the strongest winds – up to 60 km/h – the aircraft was ever subjected to. Although the ground station was partially damaged by the winds, the airplane could be landed safely in autopilot assisted mode once the winds had calmed down.

T=41 hours. Drawing circles into the night using the onboard position indicator lights.

T=41 hours. Drawing circles in the night sky using the onboard position indicator lights.

T=81 hours. Thunderstorm clouds and winds up to 60 km/h make landing the AS-2 challenging.

T=81 hours. Thunderstorm clouds and winds up to 60 km/h make landing the AS-2 a challenge.

T= 81.5 hours. Landing.

T= 81.5 hours. Landing.

T=81.5 hours. A happy AtlantikSolar team after a record flight.

T=81.5 hours. A happy AtlantikSolar team after a record flight.

Future work

Having demonstrated the multi-day endurance capability of the bare UAV platform, the AtlantikSolar UAV project will now focus on extended endurance flights with payloads including optical and infrared cameras, and atmospheric sensors. These payloads will also be carried during a long-endurance and long-distance mission of more than 12 hours and 400km that is planned for later this year in the Brazilian rain forest.

Further information

Detailed design and technical information on the UAV platform can be found in “Oettershagen P, Melzer A, Mantel T, Rudin K, Lotz R, Siebenmann D, Leutenegger S, Alexis K, Siegwart R (2015), A Solar-Powered Hand-Launchable UAV for Low-Altitude Multi-Day Continuous Flight. In: IEEE International Conference on Robotics and Automation (ICRA)”

Acknowledgements

This research was funded through ETH Zurich’s internal resources, private supporters, and the European Union FP7 Search-And-Rescue research projects ICARUS and SHERPA . In addition, multiple project partners and collaborators have contributed towards making this important milestone possible, and we’d like to thank all of them for their various and ongoing support. Finally, we are grateful towards the Rafz model aeroplane club for providing the airfield and the Aero Club Asas da Planície (Portugal) for providing the backup airfield!

Pilots: Rainer Lotz, Adrian Eggenberger, Philipp Oettershagen, Bartosz Wawrzacz. Development and Operations Team (Autonomous Systems Lab): Philipp Oettershagen, Rainer Lotz, Amir Melzer, Thomas Mantel, Bartosz Wawrzacz, Konrad Rudin, Thomas Stastny, Raphael Schranz, Jan Steger, Lukas Wirth, Dieter Siebenmann, Dr. Stefan Leutenegger, Dr. Kostas Alexis, Prof. Dr. Roland Siegwart.



tags: , , , , ,


Philipp Oettershagen is a research assistant and PhD student at ETH Zurich’s Autonomous Systems Lab (ASL).
Philipp Oettershagen is a research assistant and PhD student at ETH Zurich’s Autonomous Systems Lab (ASL).


Subscribe to Robohub newsletter on substack



Related posts :

Generative AI improves a wireless vision system that sees through obstructions

  08 Apr 2026
With this new technique, a robot could more accurately detect hidden objects or understand an indoor scene using reflected Wi-Fi signals.

Resource-constrained image generation and visual understanding: an interview with Aniket Roy

  07 Apr 2026
Aniket tells us about his research exploring how modern generative models can be adapted to operate efficiently while maintaining strong performance.

Back to school: robots learn from factory workers

  02 Apr 2026
A Czech startup is making factory automation easier by letting workers teach robots new tasks through simple demonstrations instead of complex coding.

Resource-sharing boosts robotic resilience

  31 Mar 2026
When a modular robot shares power, sensing, and communication resources among its individual units, it is significantly more resistant to failure than traditional robotic systems.

Robot Talk Episode 150 – House building robots, with Vikas Enti

  27 Mar 2026
In the latest episode of the Robot Talk podcast, Claire chatted to Vikas Enti from Reframe Systems about using robotics and automation to build climate-resilient, high-performance homes.

A history of RoboCup with Manuela Veloso

and   24 Mar 2026
Find out how RoboCup got started and how the competition has evolved, from one of the co-founders.

Robot Talk Episode 149 – Robot safety and security, with Krystal Mattich

  20 Mar 2026
In the latest episode of the Robot Talk podcast, Claire chatted to Krystal Mattich from Brain Corp about trustworthy autonomous robots in public spaces.

A multi-armed robot for assisting with agricultural tasks

  18 Mar 2026
How can a robot safely manipulate branches to reveal hidden flowers while remaining aware of interaction forces and minimizing damage?



Robohub is supported by:


Subscribe to Robohub newsletter on substack




 















©2026.02 - Association for the Understanding of Artificial Intelligence