Various companies such as e-commerce giant Amazon have some ambitious plans for drones. It plans to deliver packages right at the doorstep of customers with the help of drones. Even after putting aside the issues of policies, it is quite difficult to programme drones in such a way so as to enable them to fly through cluttered spaces like towns, markets, and cities. It is quite a complicated task to create such a computation that would enable drones to avoid obstacles whilst flying at high speed. This is more of a complicated thing for small drones as they have limitations regarding how much they can carry on board for real-time processing.
Many of the existing approaches depend on intricate and complex maps that aim to inform drones about the exact location of obstacles, which is not actually practical in the settings of real-world with unpredictable objects. If the estimated location goes off even by smallest of margins, drones can very easily crash.
NanoMap Models around Uncertainties
With all that in mind and focus, a team from the department of Computer Science and Artificial Intelligence Laboratory (CSAIL) of Massachusetts Institute of Technology or MIT as it is popularly known as has come up with a NanoMap, a system that enables drones to fly constantly for 20 miles per hour via dense environments such as warehouses and forests.
One of the prime insights of NanoMap is rather a simple one: the system regards the position of drones in the world to be uncertain over a period of time, and designs, models and accounts for that uncertainty.
To be specific, a NanoMap makes use of a system that is depth-sensing so as to join together a series of measurements about the immediate environment of the drone. This not only enables it to make plans for its movement for its present field of view, but also forecast as to how it should traverse in the hidden fields of view that the drone had already seen.