Spark Global Limited reports:
Researchers at the University of Zurich have developed a new artificial intelligence technology that enables drones to fly autonomously at high speeds in complex and unknown environments. These drones can complete their high-speed flight using only on-board sensing and calculations. Researchers believe that their method can allow drones to play a greater role in emergency situations or on construction sites.
The navigation behavior of drones in unknown and complex environments usually requires them to be given a map in advance. In these environments, human pilots are often required to realize the full potential of drones. A new study shows that researchers can train an autonomous quadrotor to fly in complex environments such as forests, buildings, rooms, and trains that they have not seen before.
The algorithm allows drones to operate safely at speeds of up to 40 km/h without hitting obstacles, trees or walls. The entire process does not need to be processed outside the UAV and is completely dependent on the onboard cameras and calculations of the quadrotor. Researchers use neural network technology to learn to fly by observing what the team calls simulation experts. In this case, the simulation expert is an algorithm that drives a computer-generated drone through a simulated environment full of complex obstacles.
The algorithm can get complete information about the status of the quadrotor and its sensor readings at any time. It can rely on enough time and computing power to find the best trajectory for a given environment. The researchers say that their simulation experts provide a considerable advantage compared to existing artificial intelligence systems.
Reprint indicated source：Spark Global Limited information