Tuesday, June 17, 2014

Techy Tuesday - Meet the Horsefly

Jeff Bezos made headlines last year on the black Friday weekend when he talked about Amazon looking into using drones to deliver packages.  I had written this off as a publicity stunt, after trying to bound the problem in my head.  How long could a drone fly?  How much weight could it lift?  How many local warehouses would there need to be if a drone carrying my five pound package is to deliver it to my house and return to base in its 30 minute battery life?  They'd need more warehouses than electric power substations. 

Whenever I conclude "I just can't imagine that working" I always have to add "but maybe I don't have a good imagination".  Meet the Horsefly Optocopter, from research collaboration between the University of Cincinnati and AMP Electric Vehicles.   
The newly designed, autonomous unmanned aerial vehicle (UAV) was developed to work in tandem with AMP's delivery trucks, creating a safe, fast and never-before-seen method of delivering goods.

Steve Burns, CEO of AMP, explains the process like this: The HorseFly will be positioned atop a delivery truck, awaiting a package from the driver. When loaded, the HorseFly will scan the barcode on the package, determine the path to the delivery address via GPS and fly away – completely self-guided – to the appropriate destination. Meanwhile, the delivery truck will continue on its rounds. After successful delivery, the HorseFly will zoom back to the truck for its next delivery run and, if needed, a roughly two-minute wireless recharge.

"Our premise with HorseFly is that the HorseFly sticks close to the horse," Burns says. "If required, the HorseFly will wirelessly recharge from the large battery in the WorkHorse truck. The fact that the delivery trucks are sufficiently scattered within almost any region during the day makes for short flights, as opposed to flying from the warehouse for each delivery."
Ph.D. student Wei Wei, lead researcher, alongside the Horsefly.  (Photo from Gizmag and UC)

The UC Press Release continues:
Key to that success and a primary reason AMP teamed with UC has been the researchers' ability to make the HorseFly safe and resilient. In addition to the sophisticated autonomous controller system, the HorseFly will have multiple built-in hardware redundancies (rotors, onboard computers, battery packs). So if, for example, multiple rotors were to fail, the HorseFly and its payload still could be retrieved safely.

"An important part of the HorseFly project is that we make a vehicle that will not drop out of the sky," Burns says. "This is the particular point that UC specializes in and where we are relying on their expertise to help us build such a safe and resilient craft."
I'm suitably embarrassed about saying "that's impossible.  It'll never work".  You just need to think about the problem differently.  Instead of centralized warehouses, decentralized delivery truck/charging stations - essentially mobile warehouses.  The drones don't deliver everything, just the things they're good for. 

Bezos' dream may be closer at hand than we thought.


  1. Once again, decentralization wins over central control.

    Drones may work from Amazon warehouses scattered across the country; they probably will work better operating from deivery trucks that number 5-10X the warehouses, and UPS/FedEx/DHL/etc. can spread cost over all deliveries, and 2-5 drones/truck offer options. A truck could become a portable mobile delivery system covering several blocks with drones, then move a mile or two and do it again. There will certainly be days where weather precludes drone use, so then the delivery truck operates as they always have.

    (On a side note, I've long been curious why delivery services don't partially operate 2200-0500 when traffic is much lighter, allowing faster travel, instead of 0800-1700. Commercial establishments are closed 2300-0500, but my front porch is in the same place at midnight as it is at noon.)

    And, speaking of decentralization, 50 states almost certainly work better than one huge federal government.

  2. SiG - Logistics is king, I guess. I always enjoy seeing someone solve a problem by finding a unique solution within the limits of their tools. I guess that's why MacGyver was so popular.

    Alien - Maybe they don't deliver at night because people creeping onto porches at oh-dark-thirty make homeowners wake up with firearms in their hands. ;-) Your comment kind of reminds me of the old joke about the engineer's comment about the blind golfers - http://www.aussiegolfer.net/golf-joke-playing-blind/

  3. Weetabix - first time I heard that joke, I thought it wasn't a joke but just a story. It's one of my favorite engineer jokes.

  4. I'll agree with you that it is a method of making that type of deliveries much more efficient and potentially feasible, but there are still a number of issues that have to be worked out before it can realistically be considered ready for use.
    Ignoring legal/ regulatory issues such as FAA authorization and insurance/ licensing requirements.
    1. I have yet to see any explanation of how the UAV will find its delivery site.
    Is it going to have a high accuracy GPS and a highly accurate delivery location (yes, there are commercial units that can do sub foot accuracy - but it takes time to get it or costs alot to do it quickly), or is there going to be a camera or lidar based delivery point recognition system, maybe the recipient prints out a pattern the UAV looks for?
    2. What is the useful load of such an aircraft? it strikes me that the best use of such a device is to deliver compact high value items.
    3. What is the effective range of the device? (range and payload are a tradeoff)
    4. The driver is mentioned as giving it a package; surely an automated loading method would be better.
    5. It seems to me the best payloads to deliver would be consistent items on a regular route - a modern version of the milkman, newspaper boy, or bread truck.
    Final point: I see lots of talk about small quad (or more) copters doing various tasks - and all the demos are across an area no large than parking lot; you might want to think about the reasons that approved/ authorized 'real world' uses such as military, SAR. etc use traditional airplane structures instead.