A ground-breaking application of autonomy is ready to accelerate the public acceptance and commercial viability of uncrewed aerial vehicles (UAVs). The collaborative autonomous drone fleets created by Cambridge Consultants offer a new generation of UAVs with pilots on the loop, not in the loop.

The efficiency and safety of our collaborative fleet hinges on the autonomous capabilities enabled by our reinforcement learning algorithms. With a commitment to transparency and trust, we put together a comprehensive AI assurance methodology that was considered at every stage of this development to ensure any safety and ethical concerns were met.

Indeed, our live demo has showcased how effectively these fleets can work together to unlock commercial, safety, efficiency and cost advantages for a range of use cases. But inevitably the shift from simulated environments to real-world scenarios still presents many complexities left to be navigated. For us as technology consultants, tackling these complexities is the next step in our work as we help ambitious companies develop the technologies and platforms to deliver autonomous UAV services for complex operations.

Use cases of the collaborative drone fleet
One promising use case is infrastructure inspection, including railway lines, pipelines, electricity power lines or offshore windfarms. The inspection of such essential infrastructure is both dangerous and costly, though the price of not ensuring they’re fully operational and safe can be even higher.

In one scenario, the drone fleet inspects a potentially dangerous railway tunnel for faults. 

Rail infrastructure inspection use case from Cambridge Consultants on Vimeo.

Working together, the autonomous fleet can tackle this hazardous and complex task with each fleet member taking on a different role. The fleet checks for hot spots in electrical panels or rail cracks, all while streaming information in real time to human inspectors who can adjust the task accordingly. Meanwhile, the UAVs can locally store larger data sets, such as 4D image captures, that can then be accessed after the mission. 

Collaborative autonomous fleets could also provide transformative insights for smart farming, identifying undernourished or infested crops and treating them immediately to minimize loss. Meanwhile, soil samples can be taken, irrigation systems checked, and the overall health and yield of the crop predicted with greater accuracy. 

Smart farming use case from Cambridge Consultants on Vimeo.

Similarly, in forestry applications, UAS can unlock illuminating data insights about the forest, helping to manage forest estates in support of net zero goals and to validate the resources associated with carbon credits.

In each of these use cases, the value is not about the drones themselves, but what they enable for our clients: enhanced safety, reduced loss, increased profits, and the capture of essential data in demanding environments. These data insights, when harnessed effectively, can optimize operational efficiency and success for a range of scenarios.

Next steps in unlocking operational value 
The potential of this technology is clear, but what else is needed to perfect the fleet for the myriad of real-world scenarios it will come up against? In my view, there are four central areas that still require exploration:

1.       Moving from sim2real
How to bridge the gap between simulated environments and real-world implementation, moving from "sim2real", will be vital. We test using 2D simulations to learn how the drones resolve tasks and develop algorithms that ensure maximum safety and effectiveness. Our live demo gives a glimpse of how the algorithm works in a real environment – but now it's time to move beyond AI algorithms, reinforcement models, and 3D simulations to the deployment of actual drones equipped with authentic software, communications, positioning and sensing capabilities.

2.       Anomaly detection for greater efficiency
Another challenge lies in the resources it takes to process large data sets. Take our rail inspection scenario. There are hundreds of miles worth of rail to inspect. Processing all that data takes time and money, with most data offering limited insights beyond the track being in good condition. Our solution would be on-drone anomaly detection. By enabling drones to identify issues in real-time, operational efficiency can be streamlined to allow focused processing specifically where anomalies are detected. 

3.       Blended communication 
The dynamic and challenging nature of drone flights necessitates adaptive communication strategies. For this, a blended communication model would be best. By incorporating Wi-Fi for local take-off and landing, satellite connectivity in remote areas and 5G in busy urban areas, the drone's comms can adapt to ensure seamless communication in diverse operational environments. 

4.       Utilizing 5G private networks
Leading on from this blended communication approach, in environments where public network providers may be absent, deploying a 5G private network could fill in the gap to ensure communication remains seamless. This tailored infrastructure ensures reliable connectivity, addressing the unique challenges of specific operational locales. For example, if the drones were navigating a tunnel away from the public network, they could use a private network as an antenna payload to relay comms to one another and keep the task on track without losing connection. 

Moving beyond the confines of controlled demonstrations, we are committed to tackling these challenges head-on. As we transition the fleet into real-world environments, our focus remains steadfast on harnessing this groundbreaking technology to enhance safety, efficiency and operational excellence for clients looking to develop autonomous UAV technology with a model of AI assurance at its core. To find out more about this exciting new generation of automated, collaborative UAV fleets and what they can do for your operation, download our recent whitepaper – Collaborative autonomous drone fleets for next level UAS operations – and reach out to continue the conversation.

About the author

Martin Cookson, Director of Digital Service Innovation
Martin works with clients to help realize their ambition in developing digital services encompassing AI, UX, web 3, cloud computing and network native. He has 30+ years’ experience working in innovation with international clients and applying new technologies to bring new services live. Martin holds an MSc in Software Engineering.

Connect with Martin

Contact Martin