TL;DR
Researchers at the Alan Turing Institute have created an AI model that enables autonomous drone teams to independently redistribute tasks and reroute in real-time during missions. The system runs on compact hardware like Raspberry Pi, processing replanning decisions in hundredths of a second, allowing robotic teams to adapt even when communication with mission control is unavailable.
Autonomous Decision-Making in High-Risk Environments
Teams of autonomous drones, submarines and satellites deployed in hazardous environments face a critical challenge: maintaining mission effectiveness when unexpected situations disrupt operations and communication links to human operators fail. Researchers from the Alan Turing Institute have developed an AI model that addresses this limitation by giving robotic teams the ability to take charge of their own planning.
The research, published in Robotics and Automation Letters, introduces a highly efficient AI model capable of redistributing tasks and rerouting robots in real time. The system’s efficiency is particularly notable—it can replan during ongoing missions in just one hundredth of a second on low-power compact computers such as Raspberry Pi devices, which can be mounted onboard every robot in a team.
“Multi-robot teams have huge potential to support a diverse range of applications, from search and rescue to environmental monitoring,” said Elim Kwan, lead author and Research Engineer at the Alan Turing Institute. “Our new model will improve the resilience of these systems.”
The practical implications are significant for emergency response scenarios. Following severe flooding in a coastal city, for example, a coast guard could deploy drone teams to assess navigable streets. If a drone were damaged during the mission, the AI model would enable the remaining drones to readjust their routes immediately without compromising the mission or losing critical data.
Looking Forward
The advancement in autonomous planning capabilities positions multi-robot teams for more reliable deployment in challenging operational environments. Dr Richard Walters, Lead Research Data Scientist in DARe (the Defence Artificial Intelligence Research Centre) at the Alan Turing Institute, noted that “by enabling multi-robot teams to replan mid-mission, we can make better use of them in the risky challenging environments in which they are most needed.”
The combination of edge computing capabilities and rapid replanning creates opportunities for expanded applications in deep-sea exploration, volcanic monitoring, space operations and other scenarios where autonomous adaptation is essential. The ability to run sophisticated AI models on compact, low-power hardware represents a significant step toward truly independent robotic teams capable of sustained operations in environments where human intervention is impractical or impossible.
Source: The Alan Turing Institute