Automating Everything - The Basic Delivery in Detail - Part 3

In response to a customer’s order, a retailer transmits a request to their nearest store or warehouse which has the goods and which has been retrofitted for automated loading. A shelf₁ holding one boxed item₂ is remotely signaled and tips up the section the item is sitting on₃ (via an electric motor₄) and allows it to slide back through a soft flap₅ at the back of the shelf and onto the upper or lower conveyer belt₆ (similar to a checkout counter conveyer belt) to slide to the back of the store for loading. As the box reaches the end of the belt, a set of laser scanners₇ turn on as it passes over them and scan it on all four sides, confirming the product and quantity. Another item₈ hanging from a metal rod₉ slides back as another motor₁₀ tips up the rod. For both types of products, the back flap on the shelf only allows one item through at a time and effectively counts them off as each swing of the flap signals an electronic counter to make sure they equal the numbers of items being sourced from the store, a stock check also performed by the scanners. The products can go from where they collect at the end of the conveyor belt to a pre-programmed cart₁₁ heading to the loading docks or be manually retrieved by a store employee₁₂. If loaded onto the cart, an electric motor₁₃ will tilt the platform₁₄ they are waiting on and tip the products down a funneling chute (much like a grain chute)₁₅ into the open cargo space₁₆ of the cart.

  

Alternatively, a humanoid robot₁₇ or a robotic cart with attached arms₁₈ heads to the location of the product, retrieves it, takes it to the location of a truck or loading dock, passes it through a scanner₁₉ and puts it either on the truck or in a loading bin where the product will wait until it is ready to be loaded. (If the store is being designed from scratch for this purpose, the conveyer belt could simply tilt downward and pass beneath the floor on its way to the loading docks, but this is not practical given the existing construction of most retail outlets.)

An automated delivery truck₁ pulls up and lowers a ramp₂ to unload an automated delivery cart₃. The cart is already loaded upon arrival by a version of the shelf-clearing technique used in a store. In this case, the shelf section or sections₄ holding the product tip forward and let it slide forward onto the slide₅ of a chute (much like a funneling grain chute ) that funnels₆ the product onto the load-carrying section₇ of the cart. Items on the upper shelves₈ have a minimum weight which causes the upper, counterweighted₉ funneling chutes to tip slightly themselves to put the funnel’s end squarely upon the cart. Items set on the floor₁₀ are raised by electric motor₁₁ on the individual, short column they sit on which is normally retracted into the floor to a sufficient height to tip onto the lowest slide.₁₂ Internal cameras allow overseers to check on any error messages or other issues that might arise inside the truck.

 

 

The cart₁ rolls up to the door₂ of a house with no stairs. Meanwhile, a remote delivery overseer₃ watches several ongoing deliveries at once, prepared to intervene, abort or cede control to another unoccupied specialist at the touch of a button₄. An automatic text goes to the customer’s phone, if requested during the order, to note the delivery is happening now.

The delivery cart halts in front of the door. It generates a directional, pre-recorded doorbell from a speaker₅ to alert the occupant. After a pause, if that elicits no response, the cart follows up with a directional knocking sound. If there is still no response, it can extend a rod tipped with a foam-rubber end₆ and tap the door three times in a clear but polite knock.

When the customer answers, the delivery cart can uncover₇ the signature pad₈ and raise a short column₉ with an electric motor so the customer can sign, if necessary, with the attached stylus. If payment is required, say for a phone ordered pizza, this column will also include the card reader₁₀, cash input port₁₁, change dispenser₁₂ and receipt printout₁₃ as well.

 

Cutaway View

Once the delivery is ready to be handed off, the cart can release the product by partially retracting₁₄ the clear plastic dome₁₅ that covers it back into the vehicle’s interior. If the retailer wants the cart to be able to take a product indoors for a customer but expects to be delivering in areas where it might track in dirt or debris, a six-wheeled version₁₆ of the same cart can move each pair of wheels in turn over a doormat and spin them briefly but rapidly while locking all other wheels in place. If further stability and leverage are required, lowering a pair of rubber tipped shafts₁₇ at the front of the vehicle while spinning the wheels will let the cart brace itself while lifting the front end fractionally to the let the wheels spin without carrying the machine’s weight. A pair at the rear₁₈ can serve the same purpose for the back wheels, and both pairs can be deployed while cleaning the center set of wheels.

If a delivery requires more complex actions than an automated cart can provide, such as removing an old refrigerator and plugging in a new one, or installing a dishwasher, or simply navigating a flight of stairs, humanoid robots can take over. Normally, deliveries will be allocated based on the level of difficulty anticipated but they can be reassigned as needed. If the alternate but more costly cart design incorporating remotely controllable robotic arms₁₉ is widespread in making deliveries for a company, a cart could simply place a lightweight ramp₂₀ that braces itself₂₁ against the ground and the stairs and roll up it. Another option is available if the larger basic cart is modified to provide it. A smaller cart₂₂ could load onto the main cart₂₃ (while that larger cart’s covering dome was mostly retracted₂₄), and the primary cart could extend a lightweight ramp₂₅ for its rider which would drive across it to the open door. As this ramp extended, retractable support rods₂₆ at the front of the cart (two to lift and balance each half of ramp) would extend to brace it and then lower or rise as necessary before locking in place to provide the smaller riding cart with one of a number of angles so that it could cross to a deck, porch or doorway either below, above or level with its mounted position. These braces would be tipped with rollers₂₇ that would support the ramp as it was first extending. The rollers would in turn retract₂₈ into the rods when the ramp was fully extended, allowing it, being anchored at its rear by a hinge₂₉, to drop down onto the broad rubber-coated tops of the braces.

But if the augmented cart and ramp are unavailable, or more complex actions are required than mounting a few exterior stairs to reach a doorway, humanoid robots can be deployed.

When one or more humanoid robots₁ are needed, they will emerge from their normal storage location, the driver and passenger seats of the truck. Remote overseers trained in their control will use haptic gloves₂ to take direct control over their hand and arm motions and watch the delivery from their robot’s perspective (through cameras₃ in its head) using standard VR goggles₄. Because their hands will be otherwise occupied, control over how the robots walk will fall to a combination of factors. Each robot’s normal programming already covers normal walking, climbing, getting up, recovering balance and overcoming basic obstacles. Overseers can use biofeedback-based controls in a simple headband₅ to handle the basic commands of whether to move, how fast and in which direction. Because required operator input is so limited, only these most basic commands need to be transmitted, and any biofeedback system that can transmit them will suffice. (These exist and have been cheap enough to be included in some video games for years.)  Given the most advanced humanoid robots are already avoiding falls and unbalancing situations as they move, if their hesitation must be overcome by issuing an override, the overseer may do so using verbal commands.

Locating delivery locations, whether for a set address or for a delivery to a customer “on the go” will be confirmed with multiple sources of information.  Geolocation of the final delivery point can use set addresses as effective landmarks, GPS coordinates and phone-location tracking, comparative video, facial recognition, texting and a confirmation code.

A customer will place an order₁ with this system, thereby triggering several commands within the app governing the sale. The app will, of course, process the sale and confirm any payment using standard programming. But the phone app will also take a brief video scan₂ of the location where the delivery is to take place, assuming the user checks off a box or clicks a radio button₃ indicating they want the delivery to be sent to their present area (this will also be an option for other apps depending on their platform’s capabilities). For on-the-go deliveries the phone will also request that the customer allow a brief scan of their own face₄, which can be used as a retinal scan₅ if they look directly at the camera, or moving the phone past their face in an arc₆ while recording the facial image to provide a reference that includes a continuous, changing 3D image (harder to simulate). Either way, the background will also be taken in₇, if only peripherally, making it that much harder to fake the biometrics by supplying the data for a single set image of a face or retina, without including the customer’s immediate environment. This scan, whether facial or retinal, will be optional unless it becomes a standard requirement to minimize fraud. The facial scan will also reference back to previous scans of the customer’s face to help confirm identity. The background scan will reference back to previous images of the street or other location (if recorded in databases the retailer might have access to, such as Google Street Map). As quantum computing becomes both widespread and cheap, they will also take a fast Fourier transform to compare these images to one another and to the images the delivering machine will see when it delivers, as well as to previous images taken as both biometric security and further facial-recognition references. Finally, when the delivery system arrives, it can check the data provided against what it is seeing upon delivery₈, or even require a final biometric scan₉ – and either way, the final data set will be collected by a device under the company’s control, not the customer’s. The customer’s phone could also video the cart and transmit that image back in real time (as with a video chat) while the cart flashed confirmation images or codes over its payment screen.

While the truck-and-cart delivery system described above can be employed in these deliveries, if the customer clicks the option for curbside delivery, these deliveries to on-the-move customers could be distributed through much smaller vehicles. Given automated-driving programs that can handle roads and city streets, smaller vehicles such as compact cars, half-cars, aerial drones or even velomobiles or a motorcycle or moped with a side car could bring products directly to the purchaser.

One option is simply a miniature version of the truck-and-cart system operating out of a very small truck or a modified van or car. In this case the vehicle stops, unloads the cart carrying the package and delivers to the customer directly as requested. Another alternative is that the delivery vehicle stops and proffers the package through an opening in its side. One panel₁ would slide back to show the item behind a second, transparent panel₂, which would slide back₃ so the customer could take their purchase₄.

A motorcycle or moped with a side car₁ could actually have the cart as the sidecar, which would detach upon arrival₂ (with the motorcycle dropping kickstands for stability) to deliver a package away from the roadside or even inside a publicly accessible structure. In the latter case, robotic arms₃ and remote oversight₄ would be needed at first to navigate most doorways₅ (using the same haptic-gloves-and-biofeedback controls employed with humanoid robots). But the cart could be constructed to be solid enough to serve as a sidecar (with its gears in neutral, being propelled by the main vehicle) with three ruggedized wheels and yet light enough to be an acceptable visitor to an office building, front porch or mall in a way that an unridden motorcycle would not. Because this version of the cart would not require a full truck or car engine or need to support a seat and passenger, it could be much lighter than most light-weight, self-powered vehicles, such as an electric bike and rider or occupied velomobile. Also, the cart can be small enough to fit through doors, something impossible for most road-worthy, human-occupied vehicles.

Other vehicles₁ could simply tow their cart₂ and pass items down a chute₃ to it as needed at each stop (the bottom of the chute would be counterweighted₄ much like the chutes described above, so that it would drop into the open dome when reloading the cart and stay closed otherwise). While this option for reloading from the primary vehicle is also possible from the motorcycle or moped version, the easiest place to put the bulk of its cargo would be on a bin secured over its seat₅. While all of these smaller-vehicle options could be modified to maximize available cargo space, a small SUV with its rear seats folded down or preferably removed would have the most room for deliveries.

With each package sitting in its own bin in the vehicle, the system would have a record of exactly where each one was and could retrieve it and pass it by the same method to the same funneling slides used in the larger trucks discussed above. These would funnel packages into the chute and thence to the delivery cart. Once items were placed in the cart, the chute would detach while the transparent dome clamped shut and the cart would detach from the truck and go to its customer.

Security risks in locations such as DC and other sensitive areas can be easily ameliorated by looping law enforcement into certain standard sensors such as internal cameras and into supplementary ones such as chemical scanners. Depending on the sensitivity of the area and the lengths a company is willing or required to go to, there are already plenty of basic devices that can scan for threats. X-ray scanning can be incorporated along with the laser scanners that double check a product’s identity before loading on the main delivery vehicle₁, preferably augmented with automatic object recognition as that software becomes viable. Geiger counters can be included at the same point_­2 as well as inside the vehicle. Atmospheric sensors inside the truck₃ can draw in air to check for explosives and other known chemical and biological hazards. By granting access₄ to this data to relevant law enforcement in the area covered by this local delivery system, the automated system becomes no more vulnerable to being highjacked by terrorists than a conventionally manned operation.

Part 1
Part 2
Part 4