Artificial intelligence is entering a new phase. Beyond generating content, analysing information or supporting decisions, AI is increasingly being used to control machines in the physical world. This shift to physical AI requires systems that can perceive their surroundings, understand changing conditions, make real-time decisions and physically act on them.
For India, this conversation is particularly relevant. The country is seeing rapidly growing aviation and logistics demand, congested urban environments, and an increasing need for faster, more efficient delivery systems.
In this episode of the Analytics Insight Podcast, host Priya Dialani speaks with Ankit Kumar, founder and CEO of Skye Air Mobility, about what physical AI could mean for aviation, why India could become an important proving ground for autonomous flight, and what it will take to make drone logistics commercially viable at scale. Here are the key excerpts:
Skye Air Mobility was founded in 2019 to address three interconnected challenges: growing congestion, high logistics costs and sustainability. India’s transport infrastructure is not expanding at the same pace as the demand for movement of people and goods, putting pressure on conventional road-based logistics. At the same time, logistics remains a significant cost for the economy, and inefficiencies such as delays, multiple handoffs and poor route planning add to the burden.
Air pollution and climate change further underline the need for cleaner, more efficient systems. Skye Air was created to make movement faster, more efficient, cost-effective and sustainable by using autonomous aerial logistics to solve problems that affect both urban and rural India.
When AI operates in software, an incorrect recommendation can often be reviewed or corrected before it causes major harm. When AI controls a machine in the physical world, the consequences are immediate.
A drone or aircraft must perceive its environment, understand what is happening, decide what to do and execute that decision in real time. At Skye Air, physical AI is viewed as the integration of three machines: drones that transport shipments, SkyPods that receive and manage packages, and robots that complete the final delivery.
These systems must exchange data, make decisions and complete secure handoffs around a shared parcel. Human oversight remains important, but machines can increasingly handle routine decisions, allowing operations to scale beyond what manual processes can support.
In a conventional delivery model, a package passes through several people before reaching the customer. It may move through a sorting station, then a delivery agent, who travels through traffic and makes multiple stops before completing the delivery.
Each handoff adds time, cost and potential points of failure. In Skye Air’s model, a drone carries the shipment to a SkyPod located within a housing complex or another suitable location.
The pod identifies the package, checks available capacity and manages the handoff. A robot can then take the package from the pod and deliver it to the customer’s doorstep. The system makes decisions in real time, assessing package volume, pod capacity and delivery location. As delivery volumes increase, machines can process multiple variables simultaneously and respond faster than manual systems.
Safety must be a foundational principle, not an afterthought. Whether an aircraft weighs 25 kg, 50 kg or two tonnes, a failure in the air can create serious risks for people and property on the ground. Environmental conditions are unpredictable, and weather can change suddenly, exposing drones to wind, obstacles or other aircraft.
While risks cannot be eliminated entirely, systems can be designed to detect problems, respond appropriately and activate incident-management procedures. Standard operating procedures, or SOPs, are essential because aviation has historically been built around disciplined processes that maintain safety even when conditions or personnel change.
Skye Air began with dual redundancy across critical systems, including motors, controllers and batteries, and is now working towards triple-redundant systems to improve reliability and safety.
As drone operations scale, airspace will become increasingly crowded. Delivery drones, aerial filming platforms and other unmanned aircraft may operate in the same low-altitude environment. Without coordination, two drones could be assigned to the same corridor, creating a collision risk. Skye Air’s Unmanned Traffic Management, or UTM, system is designed to deconflict drone movements automatically.
It uses onboard sensors, flight-controller data and system-level logic to detect potential conflicts and determine how to avoid them. Consider two drones approaching each other at 50 km/h. A human operator cannot reliably assess the situation and intervene in the fraction of a second available.
The system needs to detect the risk, calculate an avoidance response and execute it almost instantly. As the number of flights grows, autonomous traffic-management systems will become necessary rather than optional.
India has several factors that make it well suited for this transition. It has a large and growing consumer market, rapid urbanisation, rising e-commerce activity and a strong pool of engineering talent. At the same time, infrastructure constraints create a real need for alternative mobility and logistics solutions.
India also has an opportunity beyond its metropolitan centres. Rural demand is changing as well, with consumers in smaller towns and villages increasingly expecting faster access to goods, services and communication.
Skye Air works with India Post in Himachal Pradesh and Assam, operating drone corridors for postal delivery. The network connects remote locations and helps deliver items such as pensions, letters, medicines and other essential goods. For people in areas where road connectivity is difficult, drones can significantly reduce delivery time and improve access to essential services.
Two things will determine whether autonomous aviation becomes a real industry: economics and acceptance. Indian consumers are value-conscious. They are not necessarily looking for the lowest-cost product; they want value for the price they pay. If a customer pays for a service, they expect a meaningful benefit in return.
For drone delivery, that benefit may be speed, reliability, reduced congestion or access to locations that are difficult to serve through conventional logistics. The second factor is acceptance across geographies and use cases.
A technology cannot remain a pilot project forever. It has to demonstrate consistent value across different sectors, locations and operating conditions. Efficiency and speed can eventually reduce costs. When volumes increase, fixed costs are distributed across more deliveries, making the economics stronger. But the starting point must be a clear value proposition: the system must solve a real problem better than the existing alternative.