East River Airlift

Summary

This case study proposes a supervised autonomous drone food-delivery corridor between the Pepsi-Cola Sign area in Long Island City and the Cornell Tech campus on Roosevelt Island. Rather than transporting every restaurant order via bike, moped, or car using the bridge, participating couriers drop off sealed food boxes at the color-coding drop-off station near the Pepsi-Cola sign in LIC and then wait for the food to be delivered via a pre-set East River lane and landing station in Cornell Tech.

AV Use Case

What AVs are involved?

The pilot program involves the deployment of six small-sized, electric-powered food drones. The payload size of each of the drones should remain within 4–6 pounds, with GPS positioning systems, computer vision capabilities, downward-facing sensors, geofencing system, battery status monitoring system. These drones fly under supervised autonomy conditions; meaning that although their route is predefined, the flight operator has the ability to alter, halt, or reverse the drone movement if the conditions of flight prove unsafe.

The drones do not substitute any other form of delivery except as a limited freight tool. On the one side, delivery people from third-party companies like DoorDash, UberEats, or HungryPanda deliver packed food to the landing station at the Pepsi-Cola Sign site. On the other hand, Roosevelt Island’s Cornell Tech campus provides a raised platform with a landing buffer area with QR code pick-up lockers where the students can retrieve their ordered food items.

What are they doing?

The route begins at a food drop-off and take-off station near the Pepsi-Cola Sign waterfront. Couriers arrive with packed food from nearby LIC restaurants, wait in a marked queue, scan the order, and place the meal into one of three color-coded service bays: red for DoorDash, green for Uber Eats, and yellow for HungryPanda. The station checks weight, door closure, destination, and weather status before assigning the order to the next available drone.

From there, the drone takes off vertically, climbs to a controlled low-altitude corridor over the East River, and flies north toward the Cornell Tech campus. The route avoids flying over dense sidewalk crowds, park lawns, and building entrances, using the water corridor as the main operating space. At Cornell Tech, the drone descends into a raised landing platform with a roughly one-meter-radius landing surface. The food container transfers into an automated locker or staff-accessible compartment, and the customer receives a pickup notification.

Service runs only during high-demand food windows: 11:00 AM–2:00 PM and 5:00 PM–9:00 PM. During these periods, drones depart every 5–7 minutes at peak demand. Flights pause during heavy rain, low visibility, lightning risk, or sustained winds above the operator’s threshold. The pilot is designed as a permitted demonstration rather than an unrestricted delivery network.

Why here?

This corridor works because both ends already have strong public visibility and clear urban logic. The LIC side has a dense restaurant market and many existing delivery workers, so the system does not need to invent demand. The Pepsi-Cola Sign and Gantry Plaza waterfront are already places where people stop, photograph, and watch the city; a take-off station there makes the technology legible as a civic demonstration rather than a hidden logistics system.

The Roosevelt Island side is even more important for the assignment. Cornell Tech is a controlled, innovation-oriented campus with open lawn space, waterfront visibility, and a community that is likely to understand the pilot as research infrastructure. The campus also has students, faculty, employees, event visitors, and tourists who may want food from LIC without adding more delivery traffic to the island. By locating the landing port on a defined subsection of the Cornell Tech lawn, the design can show exactly how an AV-enabled service should be absorbed into public space: with queue management, signage, accessible walking paths, a no-entry safety buffer, and a clear separation between drone operations and pedestrian circulation.

Stakeholders

Who participates?

Cornell Tech serves as the Roosevelt Island site host and research partner, providing a visible living-lab setting and helping evaluate safety, user experience, noise, and pickup behavior. The Roosevelt Island Operating Corporation coordinates island-side public-realm impacts and community communication. A drone delivery operator manages aircraft, remote supervision, maintenance, flight scheduling, battery management, and safety reporting. LIC restaurants participate by packaging eligible items in sealed drone-compatible containers, while delivery platforms such as DoorDash, Uber Eats, and HungryPanda coordinate courier handoff, order verification, and customer communication.

Public agencies are also central. The FAA governs the aviation side of the operation, especially commercial delivery and beyond-visual-line-of-sight questions. NYC agencies such as NYPD and DOT review local take-off and landing permits. NYC Parks / New York State Parks coordination would be needed on the Gantry Plaza side because the origin station sits in a highly visible public waterfront environment. Community stakeholders include Cornell Tech students and staff, Roosevelt Island residents, LIC residents, waterfront park users, tourists, delivery workers, restaurant owners, and people walking or biking near both stations.

Who is impacted?

Cornell Tech users gain faster access to LIC restaurants, but the broader public also experiences a new type of urban infrastructure. Tourists and waterfront visitors encounter the pilot as a demonstration of low-altitude logistics, not just a private convenience service. Roosevelt Island residents could benefit from less bridge-based delivery traffic if some short food trips shift from mopeds or cars to a fixed aerial corridor. LIC restaurants benefit from access to a nearby customer base across the water, especially during lunch and evening peaks.

Delivery workers are the group that needs the most careful treatment. The project should not be written as replacing couriers. Instead, couriers remain part of the chain by moving food from restaurants to the take-off station, checking order quality, and loading meals into the correct service bay. The job changes from full end-to-end delivery to first-mile handoff, station operation, maintenance, and customer-support roles. Pedestrians, park users, and campus visitors are also affected because they share space near the stations; their safety, comfort, and privacy determine whether the pilot feels acceptable or intrusive.

How does the solution use their capabilities?

The project uses Cornell Tech’s strength as a technical campus by treating the landing port as a research and demonstration site. Students and faculty can analyze operational data, user behavior, noise complaints, and equity outcomes. RIOC contributes local coordination and can require community meetings, operating-hour limits, and public reporting. Drone operators contribute aviation expertise, geofencing, remote supervision, maintenance, and compliance systems. Restaurants and platforms contribute existing demand, courier networks, menus, packaging standards, and customer apps.

The design also uses the unique geography of the site. The East River creates a short, legible crossing where a drone can avoid many street-level conflicts. LIC provides the food supply; Cornell Tech provides a controlled landing environment. The visual connection between the two waterfronts makes the system easy to understand: a person can stand on either side and understand where the drone is coming from and where it is going.

How does it address their concerns?

Pedestrian safety is handled through physical separation. The Cornell Tech landing platform is raised, painted Cornell red, and automatically shut top door ensuring no food tampering from the top. Pickup lockers and the waiting line sit outside the landing area, so customers do not stand under the drone path. The LIC station uses separate service bays and an organized courier queue so workers are not crowding around the same opening. Ground markings, stanchions, and wayfinding signs make the drone zone visually obvious.

Noise is addressed through limited operating windows, a cap on simultaneous drones, and a no-hover rule except during final landing. Privacy is addressed by limiting cameras to navigation and landing verification, blurring or discarding pedestrian imagery, and publishing a short data-retention policy. Equity is addressed by reserving part of the pilot for community-serving use, such as discounted delivery windows for Roosevelt Island residents with limited mobility, and by creating paid station-support roles rather than presenting automation as a pure labor-saving tool. Governance is addressed by requiring FAA and NYC take-off/landing approvals before operation and by publishing monthly safety, noise, and usage summaries.

Relevant Blueprints for Autonomous Urbanism

The following urban design strategies are drawn from the NACTO Blueprint for Autonomous Urbanism, 2nd Edition.

Dedicated Drone Pickup and Drop-off Zones

This applies NACTO’s “People come first in the autonomous age” principle, especially its direction that people walking, biking, rolling, and resting should receive first priority in street space and resource investments. It also draws from NACTO’s curb-access logic that future streets need clearly managed access zones rather than unmanaged pickup and drop-off activity.

The first NACTO Blueprint idea applied here is the dedicated AV pickup and drop-off zone. In this project, the idea is translated from curbside vehicles to low-altitude freight: both ends of the corridor receive purpose-built, clearly marked drone exchange zones instead of informal delivery activity scattered across sidewalks and lawns.

At LIC, the take-off station is a boxy kiosk with a broad canopy roof and three color-coded loading panels: red, green, and yellow. Couriers approach through a single queue, scan an order, place the meal into the correct compartment, and then exit without crossing the drone take-off path. The queue is organized with stanchions on the boardwalk edge so it does not spill into general pedestrian flow. A “Drone Food Drop-Off Queue Here” sign tells workers where to wait, and the station face labels each loading bay by service type.

At Cornell Tech, the receiving station uses a raised circular landing surface mounted above the lawn, roughly one meter in radius, with a red vertical body and a white canopy edge. The no-entry zone is painted on the ground and reinforced with low barriers and warning signs. Food pickup lockers face away from the drone approach path, so customers retrieve orders from the public side of the station while the drone lands on the restricted side. This design directly responds to the Blueprint logic: the AV function is given a defined place, while pedestrians keep the safest and clearest path.

Slow Zone and Shared-Space Safety

This applies NACTO’s “Design for Safety” principle, which argues that autonomous systems should be shaped by street design that protects people outside the vehicle, not only by the technology itself.

The second Blueprint idea is slow-zone and shared-space safety. The drone itself is airborne, but the stations sit inside pedestrian environments, so the design problem is not only flight; it is how people move around a new automated service.

At Cornell Tech, the landing port creates a pedestrian-priority slow zone around the station. Walking paths remain continuous across the lawn edge, but the pickup queue is shifted to a paved waiting strip so customers do not block through-movement. The station uses tactile paving at the edge of the waiting area, low lighting for evening pickup, and simple icon-based signage for accessibility. A viewing area lets tourists or visitors watch the drone operation without standing in the queue or entering the safety buffer.

At LIC, the take-off station keeps couriers, park visitors, and tourists in separate but visible zones. The courier line sits closest to the station, the general pedestrian path stays open on the boardwalk, and the Pepsi-Cola Sign remains visible as a public landmark rather than being hidden behind logistics equipment. The result is a shared public space that accommodates the drone pilot without allowing the technology to dominate the park.

Human-Scaled Freight and Managed Micro-Logistics

This applies NACTO’s “Human-Scaled Freight” and “Freight and delivery services are consolidated” guidance, which supports smaller, more efficient delivery modes and managed freight exchange instead of unmanaged delivery clutter.

The third Blueprint idea is human-scaled freight. Instead of treating automation as a replacement for street life, this proposal narrows the role of drones to a specific micro-freight gap: a short river crossing between two high-demand waterfront nodes. Large vehicles are not introduced, and sidewalk robots are not added to campus paths. The system consolidates food handoff at two fixed stations, downsizes the freight vehicle to a small electric drone, and keeps local workers in the first-mile process.

This matters because unmanaged delivery technology can quickly clutter curbs, sidewalks, and public spaces. The project avoids that problem by using a station model: couriers do not chase drones, customers do not stand in landing areas, and drones do not search for random delivery points. Every trip has a defined origin, route, landing pad, and locker pickup location.

Methods

Step 1: Site photography and location selection

• Tool: iPhone camera + site observation at Cornell Tech; online visual reference for Pepsi-Cola Sign / Gantry Plaza.
• Transformation: I photographed the Cornell Tech lawn looking toward the Queensboro Bridge to identify a landing zone that was visible, open, and close enough to the campus path to work as a pickup point. I then paired that receiving site with the Pepsi-Cola Sign area in LIC because it has restaurant access, tourist visibility, and a strong visual identity.
• Result: The final scenario became a two-node corridor: LIC as the take-off and food drop-off point, Cornell Tech as the Roosevelt Island landing and pickup point.

Step 2: GeoJSON boundary mapping

• Tool: geojson.io.
• Transformation: I used geojson.io to draw the project boundary around the Cornell Tech North Lawn landing-port area. I exported the file as a GeoJSON FeatureCollection and saved it with the required filename boundary.geojson.
• Result: The final GeoJSON file gives the case study a precise Roosevelt Island site and LIC site footprint. It shows that the main design response is localized to a specific part of Cornell Tech, while the LIC Pepsi-Cola Sign station functions as the food-origin node for the delivery corridor.

Step 3: Station form and visual precedent

• Tool: Image-generation workflow using Meituan-style drone station references from Shenzhen / China park pilots.
• Transformation: I used the references to define the station’s form: a compact kiosk body, flat canopy roof, automated food compartments, and a drone landing/take-off surface. I changed the design language for the New York sites by repainting the Cornell Tech station in Cornell red and color-coding the LIC station red, green, and yellow to show multiple delivery platforms using the same shared infrastructure.
• Result: The generated station images show a public-facing food logistics system that is recognizable as drone infrastructure but still fits the waterfront park and campus settings.

Step 4: Cornell Tech landing image generation

• Tool: ChatGPT image generation.
• Transformation: I used my Cornell Tech site photos as the base and asked the model to add a raised one-meter-radius drone landing station a few meters in front of the camera. I iterated the prompt to make the station one-third taller, add an organized waiting line, and reshape the station to match the Meituan-style kiosk references while using Cornell red.
• Result: The final image shows the Roosevelt Island receiving station with the glass Cornell Tech building, Queensboro Bridge, raised landing structure, hovering drone, and a clear waiting area.

Step 5: LIC take-off image generation

• Tool: ChatGPT image generation.
• Transformation: I generated the LIC side using the Pepsi-Cola Sign / Gantry Plaza context and the same kiosk form. I revised the image so the station reads as the take-off point, with DoorDash, Uber Eats, and HungryPanda couriers lining up and dropping sealed food into the correct color-coded station bays.
• Result: The final image shows the food origin node: couriers prepare orders at the park-side station, and a drone waits above the kiosk before crossing the river.

Step 6: Video workflow

• Tool: invideo AI.
• Transformation: The two AI-generated still images — the LIC take-off kiosk and the Cornell Tech landing station — served as visual anchors for the entire video. A production bible was built first, locking the drone identity, site-specific location references (Pepsi-Cola Sign waterfront, Queensboro Bridge aerials, Cornell Tech campus lawn), and a consistent photorealistic urban-design style. The delivery sequence was then animated as four continuous clips: a wide dolly shot of the Gantry Plaza boardwalk where three color-coded couriers queue and the HungryPanda courier loads a meal into the kiosk; a close-up of the kiosk roof as the compartment seals and the drone lifts off vertically; an aerial tracking shot following the drone over the East River with the Queensboro Bridge spanning the frame; and a descending crane shot as the drone lands on the Cornell Tech receiving station and a student scans the QR locker for pickup. A captioned hero frame closes the sequence.
• Result: The assembled video runs approximately 30 seconds and visualizes the full supervised drone delivery service flow — from restaurant handoff to student pickup — in a continuous narrative that makes the corridor proposal immediately understandable without reading the written case study. Each shot was generated from site-specific photographic references to maintain geographic accuracy and realistic urban scale.

Sources and Design Rationale

The design follows the final project requirement to combine a case-study narrative, a GeoJSON map, and generated media into a site-specific AV future for Roosevelt Island. It also follows the NACTO Blueprint’s human-centered framing: automation should be treated as a tool for safer, more efficient, and more equitable cities, not as an excuse to redesign public space around technology alone. The proposal therefore emphasizes station placement, pedestrian separation, managed freight handoff, data governance, and labor transition instead of simply celebrating faster food delivery.