Hangar X Studios
In this forward-looking episode of Hangar X, host John Ramstead sits down with Cornell PhD candidate and autonomy researcher Mehrnaz Sabet to explore one of the most critical challenges in aerospace today: scaling drone operations safely and efficiently. Rather than framing autonomy as a replacement for humans, Mehrnaz introduces a powerful paradigm—collaborative autonomy—where humans and intelligent systems learn from each other to unlock entirely new operational capabilities. From search and rescue missions to dense urban drone delivery, this conversation dives deep into the infrastructure, AI, and coordination systems needed to support millions of drones in the sky. The discussion also uncovers groundbreaking work from Project Orion (NASA-funded), real-time simulation environments that blur the line between physical and virtual testing, and the urgent need for next-generation traffic management systems. If you're building, operating, or investing in the future of autonomous aviation—this episode is essential listening EPISODE HIGHLIGHTS * The shift from “human-out-of-the-loop” to collaborative autonomy * How drones learn from human experts in high-stakes environments like search & rescue * Inside Project Orion and NASA-backed traffic management innovation * The hidden infrastructure problem blocking drone scalability * Real-time simulation: testing thousands of drones without leaving the ground * The role of synthetic data in training autonomous systems * Why interoperability and communication standards are the next big hurdle * How academia is shaping the future of airspace—years ahead of industry KEY POINTS WITH TIMESTAMPS * [00:00:00] – The origin of a new idea: applying drone coordination learnings to broader air traffic management challenges * [00:01:23] – Framing the big question: Can autonomy truly scale in aerospace? * [00:04:38] – Introducing collaborative autonomy: humans and machines working together * [00:06:27] – One-to-many operations: how a single operator can manage multiple drones * [00:07:29] – Teaching drones from human behavior in complex missions like search & rescue * [00:10:09] – The birth of Project Orion and NASA’s involvement * [00:10:44] – Realization: coordination challenges exist across all airspace, not just public safety * [00:12:57] – Building next-gen traffic management infrastructure * [00:14:19] – The testing problem: why current drone test environments fall short * [00:15:39] – Simulating high-density operations (e.g., 200 drones/km²) * [00:18:24] – Real-time simulation + physical testing = breakthrough validation approach * [00:22:25] – What is synthetic data and why it matters * [00:24:07] – Key challenges: vision-based navigation and dynamic occlusion * [00:25:54] – Cooperative perception: drones sharing information in real-time * [00:26:28] – The need for interoperability and communication standards * [00:28:42] – Academia’s role: thinking 5–10 years ahead of industry * [00:32:27] – Two priorities for scaling: infrastructure and safety GUEST BIO: MEHRNAZ SABET Mehrnaz Sabet is a PhD candidate at Cornell University, specializing in collaborative autonomy and multi-agent machine learning for drones. Her research focuses on enabling autonomous systems to operate effectively alongside humans in complex, real-world environments. She is a key contributor to Project Orion, a NASA-funded initiative aimed at developing next-generation air traffic management systems for scalable drone operations. Her work spans AI-driven autonomy, real-time simulation environments, and infrastructure design for high-density aerospace systems. NOTABLE QUOTES “You cannot have a system completely out of the loop from humans. Autonomy has to be collaborative.” “The technology is already there. What we lack is the infrastructure to scale it safely.” “If one drone cannot see an obstacle, it should still know about it—from other drones.” “We shouldn’t just build for the next five years—we should build infrastructure that scales for the future.” “Test infrastructure is underestimated—but it’s critical for safety and public trust.”
93 episodios
Comentarios
0Sé la primera persona en comentar
¡Regístrate ahora y forma parte de la comunidad de Hangar X Studios!