Makes decisions onboard
Research drones make navigation decisions onboard. MonoRace adds a competition example, but these tasks do not establish unrestricted autonomy.
Last reviewed
What we are looking for
Are decisions made onboard without live operator control?
Where this applies
Research drones, including Swift and MonoRace · bounded navigation and racing tasks
What has been observed
- Swift used onboard sensing and computation to race autonomously.UZH / Nature · Aug 30, 2023
- The ZJU research platform integrates onboard perception, localization and control.ZJU / Science Robotics · May 2022
- The same experimental drone carried its navigation computation onboard.Zhou et al. / arXiv · Mar 25, 2026
- Qamcom reports fully automatic fire detection after launch using onboard processing.Qamcom Research and Technology · Oct 2, 2026
- MonoRace raced using onboard perception and control without external motion tracking.Bahnam et al. / Science Robotics · Oct 14, 2026
What limits this finding
- The paper identifies changed appearance as a failure risk for Swift.UZH / Nature · Aug 30, 2023
What we still need to know
Independent replication of the same configuration in changing conditions.
What would change our view
A traceable replication or documented failure changes confidence.
What would be enough to confirm it
A physical test must locate decision-making onboard and identify the autonomous task.
How this finding has changed
Research drones make navigation decisions onboard. MonoRace adds a competition example, but these tasks do not establish unrestricted autonomy.
Updated assessment. New scoped evidence added; rating and confidence unchanged. Different platforms and limits remain separate.
Research drones make navigation decisions onboard, including during a rotor-failure test. These bounded tasks do not establish unrestricted autonomy.
Updated assessment. Added onboard navigation on the same research configuration used for the rotor-failure test; rating and confidence unchanged.
Swift made its own flight decisions on a trained racecourse. This does not establish wider mission autonomy.
Checked again; no change. Reviewed; the rating and confidence remain unchanged. No qualifying new evidence resolved this parameter’s gap.
Swift made its own flight decisions on a trained racecourse. This does not establish wider mission autonomy.
First review. Initial assessment; no previous cycle for comparison.