An MIT algorithm can make drones capable of incredible and never-before-seen maneuvers

Render a dronedrone capable of improbable acrobatics, without losing its trajectory or its ability to stay in theairairthis is what researchers from the Massachusetts Institute of Technology (WITH) have managed to achieve by concocting powerful algorithms. The drone in question is of the tailsitter type, i.e. a aéronefaéronef which takes off on its tail to rise vertically, then tilts horizontally to fly with its wing. This aircraft architecture is without doubt the most effective and versatile for combining the advantages of a vertical take-off/landing drone with those of an aircraft conferred by the lift of its wing.

These aircraft are well suited for parcel delivery or search and rescue operations, but the concern is that they are not necessarily easy to fly. With these algorithms, this type of aircraft can execute maneuvers such as, for example, inverted or lateral flight, while retaining the maneuverability making it possible to maintain navigation in real time.

On the program: spins, loops, rolls, climbing turns and passages of porteporte performing complex maneuvers. In other words, the device is pushed into its deepest entrenchments in its flight envelope by rapidly changing it from vertical to horizontal flight while integrating lateral and inverted maneuvers.

Perform sudden maneuvers that are impossible with this type of drone, and even synchronize them with other devices, this is what MIT researchers have managed to achieve thanks to their algorithm. © MIT

Seeking the fluidity of real-time flight

For this to work, the researchers relied on the mathematical notion of differential flatness. Used for many fields of geometry, mechanics and physiquephysique, this notion allowed them to generate fluid trajectories in real time. Without it, latency wouldn’t have created these capabilities. With such a system, aircraft could navigate using complex movements in convoluted structures or with many obstacles that could impede flight. Useful abilities for the quick search of survivors in a partially collapsed building for example. In testing, these flights were performed indoors, but the team now wants to improve the algorithm so that it can be used effectively for fully autonomous outdoor flights, with environmental conditions that affect flight dynamics.