New Lex Fridman Insight: Sertac Karaman: Robots That Fly and Robots That Drive
Sent June 11, 2026
Key Insights
- Autonomous flying for consumer drones is easier than driving, but urban deployment is complex.
- LIDAR is favored in self-driving cars for ease, but future systems may rely on cameras.
- High-speed computing limits current autonomous systems, needing new architectures.
- Optimus aims for efficient urban transport with smaller, more frequent vehicles.
- Camera-based systems may eventually replace LIDAR in autonomous vehicles.
How the conversation moved
The episode begins with Sertac Karaman discussing the relative ease of developing autonomous flying systems for consumer drones compared to autonomous driving systems. He highlights that while technical challenges exist, the real difficulty lies in integrating these systems into environments populated by humans, which requires significant advancements in algorithms and societal acceptance. Karaman points out that although autonomous flying seems more straightforward, achieving a high density of such vehicles in urban settings is unprecedented and complex.
Karaman argues that the reliance on LIDAR in autonomous vehicles is due to its ease of use, although it is more expensive than camera systems. He suggests that future autonomous vehicles might rely more on camera systems as they become cheaper and more integrated. This shift is supported by his belief that camera-based systems could eventually achieve full autonomy, challenging the current industry preference for LIDAR.
Lex Fridman does not challenge Karaman's views directly, but the conversation reveals an industry tension between the use of LIDAR and camera systems. Karaman's assertion that camera systems could replace LIDAR contradicts Elon Musk's claim that LIDAR is a crutch, which could spark debate among experts about the future path of autonomous vehicle technology.
The conversation concludes with Karaman discussing Optimus's market-based approach to autonomous vehicles. He emphasizes the potential economic benefits of deploying smaller, more efficient vehicles in urban areas, which could save billions in transportation costs. The discussion also touches on the future reliance on camera systems, leaving open the question of how quickly this transition might occur and what it means for the broader industry.
Surprising moments
In-depth
Autonomous Flying vs. Driving
- Autonomous flying is easier for consumer drones than driving.
- Urban deployment of flying vehicles is complex due to human environments.
- Machine learning is crucial for flying vehicle perception tasks.
LIDAR and Camera Systems in Autonomous Vehicles
- LIDAR is easier to use but expensive.
- Future systems may rely on cheaper camera systems.
- Camera systems could achieve full autonomy.
High-Speed Computing in Autonomous Systems
- Current CPUs can't handle real-time data needs.
- New architectures are needed for high-speed computing.
- Drone racing pushes the limits of current technology.
Market-Based Approach to Autonomous Vehicles
- Optimus aims for efficient urban transport with smaller vehicles.
- Economic impact could save billions in transport-deprived areas.
- Current shuttle experiences are inefficient and disliked.
Notable Quotes
I think that autonomous flying, just doing it for consumer drones and so on, the kinds of applications that we're looking at right now, is probably easier.
Still open
- Karaman questions how quickly a transition to camera-based systems for autonomous vehicles might occur and its implications for the industry.