Amazon delivery drone is smart enough not to land on humans

Credit: Flickr

Amazon has been working on a drone based delivery system called as Amazon Prime Air for quite sometime now. Amazon claims that it can deliver packages to customers in 30 minutes or less with these drones.

The USPTO published a patent application of Amazon on March 21st, 2019 which reveals a method for detecting obstacles in the landing area of a drone.

Amazon in its filing notes that unmanned aerial vehicles (or “UAV”) are modern, versatile machines that are currently being used in an increasing number of monitoring, surveillance and delivery operations. An unmanned aerial vehicle may be configured to operate in two or more flight modes, including a forward flight mode (or a substantially horizontal flight mode) in which the aerial vehicle travels from one point in space (e.g., a land-based point or, alternatively, a sea-based or air-based point) to another point in space by traveling over at least a portion of the surface of the Earth.

An unmanned aerial vehicle may also be configured to engage in a vertical flight mode in which the aerial vehicle travels in a vertical or substantially vertical direction from one altitude to another altitude (e.g., upward or downward, from a first point on land, on sea or in the air to a second point in the air, or vice versa) substantially normal to the surface of the Earth, or hovers (e.g., maintains a substantially constant altitude), with an insubstantial change in horizontal or lateral position over the surface of the Earth.

Amazon further notes that an unmanned aerial vehicle may be further configured to operate in both forward and vertical flight modes, e.g., in a hybrid mode in which a position of the unmanned aerial vehicle changes in both horizontal and vertical directions. Forces of lift and thrust are commonly applied to unmanned aerial vehicles using one or more propellers, or devices having blades that are mounted about a hub and joined to a shaft or other component of a prime mover, which may rotate at angular velocities of thousands of revolutions per minute during flight operations. 

UAVs are frequently equipped with one or more imaging devices such as digital cameras which may be used to aid in the guided or autonomous operation of an aerial vehicle, to determine when the aerial vehicle has arrived at or passed over a given location, or is within range of one or more structures, features, objects or humans (or other animals), to conduct surveillance or monitoring operations, or for any other purpose. 

When transitioning from a horizontal flight mode to a vertical flight mode and preparing to land, e.g., to deliver or retrieve a payload, or for any other reason, an unmanned aerial vehicle must determine whether a designated landing area for the unmanned aerial vehicle is free of obstacles.

If a landing area includes one or more obstacles, such as humans, animals, structures or debris, an unmanned aerial vehicle that attempts to land at the landing area risks causing injury to one or more humans or other animals, or damage to one or more structures, as well as the unmanned aerial vehicle itself.

Where a landing area includes one or more obstacles that are sufficiently large, or that include surfaces having one or more colors or textures that are distinct from colors or textures of surfaces of the landing area, such obstacles may be readily identified in visual imaging data captured by the unmanned aerial vehicle.

Where a landing area includes one or more obstacles that have small spatial extents, or that feature surfaces having colors or textures that are similar to or consistent with colors or textures of surfaces of the landing area, however, such obstacles may not be readily identified in one or more visual imaging data captured thereby, yet may pose substantial hazards to the safe operation and the integrity of the unmanned aerial vehicle.

Credit: The USPTO

As is shown in FIG. 1A from the patent application above, the system (#100) includes an aerial vehicle (#110) approaching a destination (#170) (e.g., a house or other dwelling, an office building, or any other facility that may receive deliveries of items) that bears a target marker (#160) on a landing surface.

A plurality of obstacles (or obstructions) (#175-1, #175-2, #175-3, #175-4, #175-5, #175-6) are present within a vicinity of the target marker (#160), including a fence (or wall) (#175-1), a tree (or other natural plant) (#175-2), one or more utility wires (#175-3), a bicycle (#175-4) (or other object), an automobile (#175-5), and a plurality of utility poles (#175-6).

The obstacles (#175-1, #175-2, #175-3, #175-4, #175-5, #175-6) may be static or dynamic in nature, e.g., fixed in their respective positions or mobile.

Credit: The USPTO

As is shown in FIG. 1B, the aerial vehicle (#110) may begin to search for obstacles at the landing area when the aerial vehicle (#110) reaches a predetermined altitude threshold.

For example, as is shown in FIG. 1B, the aerial vehicle (#110) may determine that it has reached an altitude z using the range sensor (#150) or any other device, and may compare the altitude z to a threshold that may be defined on any basis, including one or more attributes of the destination (#170) or a mission of the aerial vehicle (#110), any operational considerations of the aerial vehicle (#110), or any other basis (e.g., prevailing weather conditions near the destination #170).

Upon confirming that the altitude z of the aerial vehicle (#110) is within the predetermined altitude threshold, the aerial vehicle (#110) may begin to search for the target marker (#160) and evaluate the landing area at the destination (#170).

Drone delivery systems are not yet a reality. But, we cannot completely rule them out.

Spread the word
  • 2

2 thoughts on “Amazon delivery drone is smart enough not to land on humans

Leave a Reply

Your email address will not be published. Required fields are marked *