Drones can learn from bees

The researchers took inspiration from nature to improve the movement of robotic aircraft. Staff at Delft University of Technology and Westphalian College studied and copied the movements of small insects such as bees in order to significantly improve the flight skills of unmanned aircraft. Specialists have used methods taken from nature to determine the distance to other objects for drones and to estimate flight speed faster and more accurately. Thus, a machine learning process supported by artificial intelligence can allow robotic aircraft to land quickly and extremely accurately.

Guido de Croon, project manager at Delft University of Technology, explained that if anyone has ever observed how elegantly a bee flies from one flower to another or avoids obstacles in its path, it may have asked itself how such a small insect knows so much. fly precisely. The solution is the so-called optical flow, which allows insects to estimate well the distances to nearby obstacles. The researcher added that some may think that only biologists are interested in how bees can land on each flower and avoid obstacles. However, the growing proliferation of small-scale electronic devices and robots has led to these issues becoming increasingly important in robotics and artificial intelligence research.

But Croon said small-scale aerial robots are very limited in the sense that they can’t carry many sensors and processors. If these machines want to work on their own, just like autonomous cars, they need a whole new artificial intelligence. Christophe De Wagter, another specialist, said that estimating distances and speed based purely on optical flow has led to their unmanned aircraft being able to land much faster and in soft. The first tests were successful and not only did object recognition improve, but so did the overall speed of the robots.