Drone Based Deliveries
Unmanned Aerial Vehicles, aka Drones, can be useful for rapid delivery of goods to retail customers. People can order groceries, take-out food, prescriptions, and other supplies from nearby vendors. Instead of human delivery agents, the vendors can use drones to rapidly deliver the purchased items. Drones will be less expensive than humans and can function all hours of the day. However, drones require a delivery point at the customer's site. A customer who resides in a multi-story apartment complex may not have a convenient point where they can receive drone deliveries. Even suburban dwellers may not want deliveries dropped on their driveways or backyards. This note describes a vision for Drone based deliveries that utilizes an innovative new delivery point called the Drone Port.
Our vision is to have fully autonomous drones deliver small items to customers in densely populated metro areas. Each package would typically be limited to about 8 pounds and be smaller than a 9-inch cube. Once the customer places an order, most of the fulfillment steps can be automated. The loading of items onto a drone can be done with minimal human intervention. Drones from different vendors can file their flight paths and get clearances from an automated traffic control authority. Deliveries can be automated, and the customer need not be present.
Several drone delivery trials have been started, for example, by Amazon, Google (Wing), UPS, Zipline and Walmart. While they are a great start, they have not been widely accepted. There are several reasons for this:
1. The drones drop their cargo on the customer's front yard, where they can be stolen or damaged. None of the trials have addressed the critical issue of the delivery point, which is a key requirement if a delivery service is to be acceptable to retail customers. A few video examples illustrate this point in the Reference section below. A method for delivering packages securely to the customers’ rooms using the Drone Port is described here.
2. The goods available for delivery in each trial are limited to what is available from that one vendor, like Amazon, or Walmart. A successful trial will need to involve multiple independent vendors, including small businesses. Though Amazon does have an incredible range of products, they do have some problems, which are described below.
3. The drone operation is not autonomous. Humans are needed to load, recharge and fly the drones. Humans need to be paid, and this makes the trial wildly uneconomical. Drone operation needs to be fully autonomous. Besides being economical, autonomy makes the service more reliable and trustworthy.
4. There is a lack of standards that are needed to allow multiple drone operators to operate independently. Competition between operators will help drive down costs and improve service.
This note addresses the above deficiencies.
Every project should start by defining the goals and requirements. There are six high level goals:
Convenience is key to the success and widespread adoption of this project. Drones should be able to deliver packages conveniently to the customers. Similarly, they should be able to pick up packages without requiring much effort from retail vendors. The Drone Port is a solution that can securely deliver a package to a room in the customer’s home.
There are many parties involved in each delivery transaction and security is important to each party. The three main parties are: the customer receiving the package, the vendor sending the package, and the drone service operator.
· The customer needs to be assured that only authorized drones can enter their homes to make authorized deliveries. Unexpected deliveries may be rejected, depending on the recipient’s choice. All drones must be uniquely authenticated. Each delivery will be accompanied by a digitally signed manifest that lists the source and the package contents.
· The Vendor will need confirmation that the package was delivered to the correct recipient. Digital payments would rely on this confirmation.
· The Drone Service Operator should be assured that the drone will not be stolen or abused at any point during the delivery.
Security also includes payments and financial guarantees.
Drones need to operate in a fully autonomous mode for drone-based delivery to be economical. Paying a human to pilot a drone will not be profitable on any scale.
Besides piloting, flight planning and control should be fully autonomous. Air Route Traffic Control Centers must be automated to manage drone flight plans. Weather reports and local wind condition warnings must also be automated. Only a small number of humans should be needed for monitoring the operations and for rules enforcement.
The system should smoothly allow a very large number of simultaneous drone deliveries. Besides technology & economics, the policy for drone operations should:
· Allow local decentralized controls (like local traffic signals)
· Be based on open standards
· Reasonable regulations
· Reasonable fees and licenses
A large variety of goods should be available for delivery by drones. This means many “Small Business” vendors should participate, and this means that vendors should not have to make a large investment to participate. Small Business vendors are discussed further below.
The delivery service should be available to most residents in most areas. Delivery in densely populated metro areas will be much appreciated, and very profitable. Delivering large packages to distant locations may require larger drones.
Several aspects of the vision have been introduced as the Goals. This section describes more high level descriptions of the system, including how it will be used, the requirements and non-requirements, performance, cost, etc. They are summarized below.
About 35% of the metro population are young people between the ages of 18 to 44. This demographic has an above-average income and an inclination to spend for immediate gratification. They would seek convenient deliveries, frequently at odd hours late at night. Many may live in high rise apartments which may not be amenable to human deliveries. These people are used to the convenience of Grubhub, DoorDash, etc. and are willing to pay for the service.
The packages delivered are initially limited to about 8 lbs. and sized approximately like a 9-inch cube. This should comfortably suffice for a dinner for two, or for half a gallon of milk. This constraint is due to the limitation of the size of drones that are likely to be approved to operate in metro areas. Regulators may not approve heavier packages out of concern that there may be injuries or property damage if the packages fall accidentally. Larger drones may not fit between buildings in metro areas. Note that this limitation is subjective and may be changed.
While some current trials deliver large packages, this project is not designed to deliver a 25 lb. laser printer. Instead, it will conveniently deliver the 5 lb. replacement toner that is urgently needed late in the evening. Also, some trials use fixed wing drones that are geared towards longer delivery distances. Such drones may not be suitable for the current vision.
There will be a standard that will define several types of packages. This will allow drones from different manufacturers to be used for delivery. Each type will have a specific
1. shape and size,
2. maximum load,
3. type of contents (solid, liquid, granular, hazardous, etc.),
4. location and type of gripping points.
The Delivery time is the transit time for a package. It does not include the time needed to prepare the package contents. The average delivery time using humans will be at least 35 mins. Further, a human agent may choose to make multiple deliveries in one trip, and this may add to the delivery time latency.
In comparison, for a drone delivery, the transit time for 2 miles (at 15 mph) would be at most 10 mins. There will typically be only one package per flight and the drone will not wait for multiple packages to be prepared for delivery.
Drone delivery today is driven by separate projects run by large companies, like Amazon, Walmart, UPS, etc. While this is good for the development of technology, the type of goods that can be delivered will be limited to what each company sells. There is only so much that a customer would want from Amazon. A customer will want deliveries from multiple small businesses, like restaurants, grocery stores, smoke shops, hardware shops, dry cleaners, etc. Clearly, drone delivery trials should include a variety of such vendors. This will eventually happen, but small businesses should be included in early trials to understand their special needs.
Including small businesses in a trial has a couple of big implications:
1. Small businesses cannot afford to have their own dedicated fleet of drones. They will need the services of one or more Drone Service Operators.
2. Small businesses would typically include shops in a strip mall. They may not have much room to install a landing pad for drones. A Drone Port (described separately) would allow them to support drone deliveries in the limited space they have. A delivery order would be prepared indoors. When it is ready, a drone would swoop in, pick up the package from a Drone Port, and zip away to deliver it.
Drone Service Operators provide drone rentals much like a taxicab service. When there is a delivery request from a customer, the retail vendor would forward the delivery details to a drone service. The service would assign a suitable drone to pick up the package at the vendor and deliver it to the customer. Eventually, there will be multiple drone service operators available and the service with the best fit will be picked. This will need standards for package sizes and form factors.
Each drone service operator would deal with financing and insuring their drones. The costs can be recovered by charging vendors for each delivery. Some approximate cost estimates are shown below.
Drone service operators will be responsible only for the safe and timely delivery of the package. They will not be responsible for the quality or correctness of the contents of the package, just like the Post Office is not responsible for the contents. Customers will have to deal with the sending vendor to negotiate refunds or replacements.
Each drone service operator can negotiate service agreements with vendors that define the capabilities, price, and response times of their drones. This allows vendors to select the most suitable drone for each delivery. The drone operator can also approve and validate each receiving customer to verify that their drone will be safe while delivering to the customer site. Some customers may be blacklisted if they misbehave.
Like auto mechanics, there can be specialized Drone Mechanics that provide repair and periodic maintenance service to Drone Operators.
Like gas stations, there can be Drone Recharging services. Given that a drone may take more than an hour to recharge, the recharging facility would need a large area to allow multiple drones to charge simultaneously. A large number of drones would land and leave from here. This facility does not have to be run by the Drone Operator, just as taxicab companies do not operate gas stations.
The large companies currently running delivery trials can act as Drone service operators, Mechanics, or Rechargers by providing service to other retail vendors. It is likely that companies like DoorDash & Grubhub will also start using drones. Other potential operators include US Mail (USPS), FedEx, etc.
Though autonomous piloting is technically possible today, there are government regulations that currently block deployment in urban areas. The regulations are evolving, and it will take a few years for autonomous deliveries to be ubiquitous. However, trials are needed today to help shape the regulations.
Besides Federal regulations, the local government will also need to permit autonomous drones.
Besides supporting autonomous flights, drones will also need to pick up and drop off packages without human intervention. Each drone model should be able to support one or more package standards. While requesting a drone, a vendor would specify the package type and expect to be assigned a suitable drone.
Recharging drones should also be autonomous. It should not require a human to manually plug in the charger. There can be multiple standards for charging, including wireless inductive methods.
A drone may optionally carry more than one package and may be loaded with a package when it is being charged.
A key part of the vision is a mechanism that allows convenient package deliveries and pick-ups. A Drone Port is such a mechanism that can be installed in a window of an existing building. It will allow drones to safely deliver packages to customers. After delivery, the Drone Port will move the package indoors to protect it from the weather and from passing miscreants. The Drone Port is detailed here.
The architecture for autonomous delivery depends on many components which currently do not exist. Open standards would allow independent participants to develop components which can interoperate. ISO has a committee, ISO/TC20/SC16 for Unmanned aircraft systems, though the IETF may also be a suitable forum. The standards can leverage the experience gathered over years of non-drone operations. We would need:
1. A standard protocol between the Drone Ports and the package originator for authorizing delivery to a recipient and acknowledging the delivery. This prevents 'junk mail' deliveries and provides a delivery confirmation to the sender.
2. A standard protocol between drones and the airspace authority to mediate the flight paths of multiple independent drones in an airspace, analogous to today's Air Traffic Control. It can borrow best practices from today’s commercial ATC. For example: ATIS, Airport Tower procedures, RNAV, ILS, etc.
3. A standard for software that can autonomously fly a drone along a planned flight path. This is required for economical and scalable operations. Though the software itself may be proprietary, this standard would specify minimum requirements.
4. A standard protocol between Drone Ports and drones to provide terminal guidance at the landing points. This will help navigate around trees, wires, and grumpy neighbors. This is analogous to an airport's Tower & Departure controls. It also triggers the opening of the Drone Port when the drone is approaching.
5. A standard for package form factors. Just as cargo shipping containers are standardized, this standard would specify gripping points, dimensions, weight limits, content restrictions, etc.
6. A standard for autonomous charging of drones. This would specify the charging port location, adapter, voltage and current. Third party charging would need accounts for billing and payments.
Today, the typical human delivery charge is estimated at $11. A human is paid $20/hour plus $10/hour for fuel costs and can make 4 deliveries per hour. This results in a cost of $7.50 ($30 / 4). The average value of a package that is delivered is estimated to be about $25. The cost to the customer would include a 15% tip ($3.75), adding to a delivery cost of about $11 (about 45%.) These costs have been accepted by most customers. Note that part of the cost may be hidden by inflating the price of menu items, vendor contributions, subsidies, etc.
Drone deliveries do not involve wages, tips, etc. Instead, there is a huge investment cost that needs to be amortized. The starting costs for 200 drones and charging infrastructure are estimated at $500K. Initial software development is estimated at $1M. A drone can make one delivery per hour, allowing for recharging times. A fleet of about 200 drones can make 750 deliveries per day. Allow $250,000 for annual fleet maintenance. The first-year costs add up to about $1,750,000. There may be government subsidies.
Assume that the charge for a drone delivery is $10. The marginal operational cost for a 4-mile round trip would be about $1, leaving $9 to pay off start-up costs. The daily profit would be $6,750, and the investment could be profitable in less than 1 year. A better estimate would factor in the distance, weight, value, and time of the delivery. However, people would be willing to pay for quicker delivery, convenience, and the value of a delivery untainted by unknown human intermediaries. Note that drone cost estimates vary widely in available references.
The delivery charge ($10) will effectively be borne by the vendor, who may pass the cost along to the customer. The initial investment cost would be borne by the Drone Service Operators.
As a rough estimate, a typical Drone Port can retail for about $1,200. The charge for a professional installation on an upper floor could be about $500, resulting in a total cost of less than $2,000. A regular window replacement of comparable size would cost around $1,200. Replacing a window just for the convenience of receiving drone deliveries seems expensive but may be justified as follows. Besides the end customer, each installed Drone Port will benefit several other parties. There will be multiple local retail vendors who can now deliver goods to that customer. Drone operators will see an increase in business. Businesses that provide drone repairs and maintenance will prosper. The governments will see an overall increase in tax revenue. There could be an initial government subsidy or rebate to kick-start this industry.
The following summarizes the steps for a typical delivery. Many details and checks are omitted. The Drone Port is only a small part of the sequence.
1. Customer uses an online app to place an order with a participating vendor. Information sent includes:
a. Order and Payment information,
b. Customer’s Drone Port information, which includes GPS location, approach path warnings, etc.,
c. Security ID and certificates,
d. Available delivery times, etc.
e. Vendor validates the order (including payment), estimates the pickup time, and contacts a Drone provider to request a drone. Multiple drone providers may be contacted to get the best delivery price and pickup time. Information sent includes:
i. Vendor’s and Customer’s Drone Port information, which includes GPS location, approach path warnings, etc.
ii. Available delivery time windows, priority, package shape/type, size & weight, contents, etc.
iii. Vendor’s account and billing details.
2. Drone provider selects a suitable drone, verifies the availability of flight paths, and responds with a drone delivery quote.
a. The price, type of drone and ETA are included.
3. Vendor accepts one of the quotes and responds to the providers.
4. The selected drone provider charges the vendor, files the flight plans, and schedules the previously selected drone.
a. The response to the vendor includes the drone ID and security information to authenticate it.
b. There may be up to 3 flight plans filed, one for each leg.
5. Vendor sends the customer a confirmation, a payment receipt, along with:
a. a delivery manifest,
b. drone ID and authenticating information,
c. and an ETA.
6. Vendor assembles the order in a package acceptable to the drone and places it on their Drone Port for pickup.
7. The drone follows the filed flight path to the vendor’s location.
8. When the drone arrives, it is validated, and the package is moved outdoors to be picked up by the drone.
9. The drone follows the filed flight path to the customer’s location.
10.When the drone approaches, the Drone Port:
a. Authenticates the drone,
b. checks the delivery manifest to verify the sending vendor and the package contents,
c. opens the outer sash,
d. and may provide terminal arrival guidance to the drone.
11.The drone deposits the package and then follows the filed flight path to its next pickup point, or to its home base.
a. The Drone Port may provide departure guidance to the drone.
12.The package is moved indoors by the Drone Port, and the delivery is completed.
a. An optional mechanism can unload the package from the Drone Port’s inner sash to prepare for another delivery.
There are several components and participants mentioned above. The following diagram roughly shows the relations between them.
GrubHub (and Uber Eats, etc.) has a similar model, but uses human delivery agents instead of drones. They are limited to food delivery, for now. Drone service will be faster, economical, convenient, and reliable for small packages. Grubhub can continue with large deliveries.
The UPS model has their trucks bring packages to a neighborhood and then uses drones for last mile delivery. The implementation does not seem complete. UPS can use the proposed model for delivering small packages to apartment dwellers.
Amazon is much more than a delivery service, they are a reseller between the customer & vendor. They are more suited for wholesalers rather than existing vendors because they process product returns, refunds, discounts, etc. Vendors are forced to pay a healthy margin to Amazon, while drone based deliveries will incur a fixed fee.
Amazon uses a hub & spoke model, which requires storing inventory at hubs. This model is suitable for swift nationwide deliveries, but not for delivering pizzas locally. Amazon can use the proposed system to improve their last mile delivery efficiency. Amazon may also participate by being a Drone Service Operator.
Drones are driving several new commercial (non-military) services. However, projects involving large-scale delivery have not been economically successful. There are several challenges, like fully autonomous flying. The lack of secure and convenient delivery points for residential customers is also critical and has not been widely recognized. If successful, Drone based deliveries would have a great impact on society.
The vision involves many participants and will take a long time to implement. There will be many detractors and politicians. Note that it took more than 10 years for the Electric Vehicle vision to be recognized, and it will take even longer for its widespread use.
While great progress has been made by the current projects, the following references show only the delivery strategy, which is relevant to this note.
2. Amazon's drop-ship technique
3. UPS delivery and launch preparation
A food delivery project:
6. Meituan delivers food in Shenzhen, China
--Raj Pejaver 720