Winged parcel delivery: Google’s way
Only a few months ago, web competitor Amazon had showcased octocopters for its futuristic Amazon Prime Air delivery service with vertical take-off and landing for parcel delivery. Now, Google’s way of implementing a drone-based delivery service would rely on a winged drone.
Under secret development for the last two years at the Google X research lab, the GPS-guided drone can follow programmed routes at altitudes varying from 50 to a hundred meters, very much like the gliding drones from Parrot’s sister company, senseFly Ltd.
It has a wing span of about 1.5m for a height of 0.8m, with four pivoting propellers that allow the drone to pitch upwards to the vertical to hover when it has reached the delivery address. Then, rather than landing, the drone opens its belly and winches the parcel down, via a sensor-laden tether that will decelerate the parcel before touch-down and release it as it reaches the ground.
This delivery method was preferred over landing as previous experiments showed that people receiving the parcel couldn’t refrain from grabbing the parcel or/and the drone, at the risk of getting injured or damaging the drone.
In the first public video of the Project Wing, MIT roboticist Nick Roy who took a two-year sabbatical to lead Project Wing admits that the company is still years from a final product, but this is the first prototype that Google wants to stand by.
This is probably not only because of logistical issues such as finding collision-free routes, the right energy optimisation and ensuing drone reliability, but also because currently no sky regulation in the world would allow flying such drones over populated areas (see Who owns the space above your garden?).
Now, Google stepping into the commercial drone game and joining the ranks of Amazon and Parrot means more weight for the lobbyists, and clearly the company hopes that once the safety and privacy issues will be dealt with, there will be room for some amendments.
In the meantime, route optimisation in a sky cluttered with self-flying drones is a big topic of research. The future of optimisation may be in the cloud where computational resources abound, before take-off.
Then lighter algorithms can take-over during the drone’s flight. Going in that sense, MIT’s Department of Aeronautics and Astronautics has been working on algorithms that allow drones to monitor their own health during long package-delivery missions, while dealing with real-world uncertainties (such as high winds, sensor measurement errors, or drops in fuel).
Rather than looking at every possible scenario, and rather than rely on drone-to-drone communication, the MIT researchers have combined a health-monitoring algorithm that enables a drone to monitor aspects of its “health” in real time (to predict its fuel level and the condition of its propellers, cameras, and other sensors throughout a mission), together with a method for a drone to efficiently compute its possible future locations offline, before it takes off.
The method, a variation of the so-called Partially Observable Markov Decision Processes (POMDP) where each decision node is represented by a probability distribution, allows the drone to accurately predict the collision and failure probabilities on different routes. This probabilistic route planning ahead of take-off is said to free up computational energy for mission-level planning in real time.
Related articles:
Who owns the space above your garden?
Amazon bets on GPS-driven drones to deliver goods on your door-step
Swarm intelligence controls robotic planes
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
