Signal 5: The Real “Unmanned” Aerial Vehicles build AI cities

Junjie Cai
Civic Analytics 2018
2 min readOct 28, 2018

At present drones are commonly more like remote-controlled aircrafts than artificial intelligence aircrafts. However, research laboratories and application industries are constantly striving to develop the AI drones. Drones will be truly automated and think like humans. And the AI drone cluster will become an important element and cornerstone for AI cities.

Microsoft and DJI have strategically cooperated to promote the combination of artificial intelligence, machine learning technology, and drones. They aim to enable enterprise users to reinvent their productivity by applying drones and smart edge cloud computing. The AI solutions can be trained in the cloud and then deployed to drones for real-time AI operations. This solves the limitations of centralized AI computing data transmission and meets the real-time requirements of flight operations so that each industry drone can have powerful AI computing capabilities.

The AI drone cluster realizes self-organizing flight and automated working, serving as the basic element of AI city. AI drone cluster forms a coherent group through self-organization. The formation of the drone group did not involve the central control system, nor did any of the drones as “commanders” or “captains” give orders. The bionic interaction enables the fleet to self-organize. From an interaction perspective, this group of drones is similar to animals in the natural world where individual behaviors trigger group behavior. They can achieve cluster intelligence in a decentralized robotic system. Each drone is only responsible for itself, and group behavior emerges spontaneously.

The combination of drone and AI technology enables the whole process automation from take-off to data analysis, which can realize the intelligence of security, inspection, urban management, agriculture, and other industries. Ultimately real AI city will be built based on the AI drone clusters.

Inspiration and background: http://robotics.sciencemag.org/content/3/20/eaat3536/tab-article-info

--

--