Autonomous Flying Cars and Geospatial Earth Models
A match made in space
The vision for Machine Learning and AI is finally expanding to include spatial and temporal data, leading to the development of new “Earth or World models.” These models promise to reshape the GIS and geospatial landscape forever, offering humanity a better way to understand, manage, and control the world.
This has sparked a debate: will a single, all-encompassing Earth model emerge, or will we see a future of specialized models tailored to specific applications? I believe the latter is the better path. The concept of one model to rule them all is fraught with risk, especially when it could influence the very fate of humanity.
Reflecting on the limitations of a single model led me to contemplate the vast concepts of time, space, and humanity. A deeper examination, supported by both philosophy and ancient wisdom, reveals a fundamental truth: everything is relative, in constant flux, and profoundly interconnected. This understanding suggests that a singular, universal model may be an impossible aspiration for humankind by design.
Earth Models
When considering humanity’s relationship with Earth, it becomes clear that while we often apply new technology to old problems, the greatest leaps in history have come from exploring new frontiers. The invention of flight serves as a prime example. While modern tools can help us find minerals, clear forests, plant crops, fight wars, locate a new store, and build data centers more efficiently, these are merely optimizations of existing workflows, assumptions, and policies. They are bound to the Earth’s surface—a living system far more complex than the human body, which itself remains largely a mystery to us.
Models designed for Earth must account for existing infrastructure. Autonomous vehicles are a prime example; their design is severely limited by the need to operate within current transportation systems, which can stifle innovation. While advanced sensors and sophisticated models offer significant progress, they have their limits. The driving environment is constantly changing, often quickly and unpredictably. The complexities are further compounded by the mix of autonomous and human drivers on the road.
Autonomous Flying Vehicles
How can we simplify autonomous vehicles? First, imagine a regulated, open space with minimal infrastructure. Next, ascend above the complexities of ground-based travel. Finally, remove human error from the equation. The answer lies in autonomous flying vehicles.
Picture your “car” as a personal device parked at your home. You simply enter it, input your destination as you would with Google Maps, and relax as it transports you to the store, work, or a friend’s house.
Unlike ground-based systems, this navigation model is significantly simpler. It only interacts with the Earth’s surface during takeoff and landing, communicating with other flying vehicles through advanced sensors and new technologies. Each vehicle functions as a mobile sensor, constantly feeding atmospheric data back to the central model. This shared, real-time information allows for immediate and dynamic adjustments across the network.
Shared autonomous vehicles could pick people up on demand and be easily lent to others, leading to reduced traffic and pollution. This is the future of transportation, but we can only achieve it by building the models to support it.
How do we build these models? What data and sensors are needed to create the core reasoning engine? How do we combine broad weather models with localized microclimate data?
The insights we’re gaining from drones are critical, but we must rapidly advance to the next level. We should view drones as test cases for autonomous cars, collaborating with geospatial experts, transportation specialists, and regulators to design the highways of the future. While companies like Joby and Archer Aviation are developing the vehicles, we must simultaneously build the comprehensive Earth Model that will support them.
What’s your take? Can you envision other “new frontier” applications powered by Earth Models?