Sertac Karaman: Robots That Fly and Robots That Drive | Lex Fridman Podcast #97 | Summary and Q&A

TL;DR

Autonomous vehicles, whether flying or driving, present different challenges, with the potential for large-scale deployments leading to significant societal impact in the near future.

Key Insights

• 😠 Large-scale deployments of autonomous vehicles face challenges related to human interaction, safety, regulations, business models, and societal impact.
• ▶️ Simulations play a critical role in training and developing safe flying robots, enabling testing in realistic environments.
• 🚙 Flying cars have the potential to revolutionize transportation, but technical feasibility and societal acceptance are important considerations.
• 🎯 Optimus Ride is targeting transportation-deprived environments with efficient, sustainable, and affordable mobility solutions.

Transcript

the following is a conversation with Suresh Carmen a professor at MIT co-founder of the autonomous vehicle company optimist ride and is one of the top roboticists in the world including robots that drive and robots that fly to me personally he has been a mentor a colleague and a friend he's one of the smartest most generous people I know so it was ... Read More

Questions & Answers

Q: What are the main challenges in deploying autonomous vehicles at a large scale?

Deploying autonomous vehicles in environments with human presence involves complex algorithms, systems, and societal considerations. Interaction with humans, safety, regulations, and business models are among the key challenges to address.

Q: How feasible are flying cars and how will they impact transportation in the future?

While flying cars have long been a dream, the technical feasibility is uncertain. However, in certain airspace, such as agile airspace, the possibilities for transportation and personal travel become more promising. Flying cars could provide faster, more efficient transportation options.

Q: How do simulations play a role in training safe and intelligent flying robots?

Simulations are critical for development and training of flying robots. Simulating interoceptive and exteroceptive sensors, as well as human behavior, helps in understanding limitations and training perception algorithms. Simulations also enable development in safe environments before real-world deployments.

Q: How do different autonomous vehicle companies, like Optimus Ride, Waymo, and Tesla, differ in their strategies and timeframes for mass deployment?

Optimus Ride focuses on geofenced environments to improve transportation and reclaim parking spaces. Waymo takes a longer-term approach with research-like projects for unrestricted movement. Tesla aims to incrementally improve its Autopilot system to achieve full automation in the near future. Timeframes for mass deployment are difficult to predict.

Q: What are the main challenges in deploying autonomous vehicles at a large scale?

Deploying autonomous vehicles in environments with human presence involves complex algorithms, systems, and societal considerations. Interaction with humans, safety, regulations, and business models are among the key challenges to address.

More Insights

• Large-scale deployments of autonomous vehicles face challenges related to human interaction, safety, regulations, business models, and societal impact.

• Simulations play a critical role in training and developing safe flying robots, enabling testing in realistic environments.

• Flying cars have the potential to revolutionize transportation, but technical feasibility and societal acceptance are important considerations.

• Optimus Ride is targeting transportation-deprived environments with efficient, sustainable, and affordable mobility solutions.

• Autonomous vehicle companies have different strategies and timeframes for mass deployment, with a focus on iterative learning and technology advancements.

Summary

Suresh Carmen, a professor at MIT and co-founder of Optimus Ride, discusses the challenges and potential of autonomous flying and driving vehicles. He explains that while autonomous flying is currently easier to achieve for consumer drones, the real benefits and challenges lie in applications like transportation logistics. He emphasizes the importance of putting robotic vehicles in human-present environments and the need for advanced algorithms, systems, business models, and societal considerations. Carmen also explores the role of simulations in developing safe and intelligent flying robots, the potential for tens of thousands of delivery drones, and the feasibility of flying cars. He discusses the different strategies and approaches taken by companies like Waymo and Tesla in the autonomous vehicle space and highlights the importance of informed public opinion in shaping the future of autonomous vehicles. Additionally, Carmen explains Optimus Ride's approach of using humans to operate autonomous vehicles in certain environments, creating efficient, safe, and sustainable transportation systems.

Questions & Answers

Q: What is the more difficult task, autonomous flying or autonomous driving?

Autonomous flying is currently easier to achieve for consumer drones, but the challenges and benefits of autonomous flying in transportation logistics make it more difficult than autonomous driving.

Q: What are the challenges of putting robotic vehicles in human-present environments?

There are several challenges, including human-robot interaction, developing fundamental algorithms and systems, designing business models, addressing architectural and planning considerations, and dealing with legal and societal issues.

Q: How likely is it that there will be tens of thousands of delivery drones in the future?

It is quite possible, especially for transportation purposes such as moving people from one location to another, like Boston to New York. Personal transport in autonomous vehicles could also become popular, with multiple people traveling together.

Q: How feasible are flying cars?

Flying cars have been a dream for many years, but technically, it is difficult to predict. However, there is potential for innovation in the agile airspace, which is high enough to prevent interference with larger aircraft but low enough to allow for utilization of the airspace.

Q: What is the role of simulations in developing safe flying robots?

Simulations are useful for both training and development. They allow for the testing and evaluation of perception algorithms, sensor simulations, and even simulating human behavior. Simulations can help build complex software and model human actions in a virtual environment.

Q: How promising is the possibility of training and developing safe flying robots in simulation?

Simulation environments can be promising for development purposes. They allow for the exploration of limitations, training of algorithms, and understanding physics and sensor interactions. However, simulating human behavior accurately remains a challenge.

Q: What are some potential societal implications of deploying large-scale autonomous vehicles?

The deployment of autonomous vehicles at a large scale raises questions about efficiency, sustainability, and livability. There is a trade-off between efficiency and sustainability, and choices need to be made regarding societal impact and livability. Designing systems that are efficient, sustainable, and acceptable to the public becomes crucial.

Q: How do strategies of companies like Waymo and Tesla differ in the autonomous vehicle space?

Waymo takes a more cautious research-oriented approach, aiming to understand the AI engine and build fully autonomous vehicles. Tesla, on the other hand, has a product-focused approach with continuous updates and a business model that appeals to customers, even if full self-driving capability is not immediately achieved.

Q: What is the concept of humans operating vehicles in the Optimus Ride approach?

Optimus Ride uses human intelligence to control multiple vehicles from a centralized location. Humans act as AV controllers, overseeing the vehicles' operations and ensuring efficiency and safety. This approach allows for high-level decision-making and coordination in transportation systems.

Q: What is Optimus Ride's viral application, and how will it connect with people in mobility?

Optimus Ride targets transportation-deprived environments or geofences where there is a lack of accessible transportation options. By providing an affordable and sustainable transportation system within such places, Optimus Ride aims to reclaim space currently used for parking and improve mobility.

Takeaways

Carmen's discussion highlights the challenges and potential of autonomous flying and driving vehicles. While autonomous flying is currently easier to achieve, the real benefits lie in applications like transportation logistics, which present complex challenges related to human-robot interaction, algorithms, systems, and societal considerations. Simulation environments play a crucial role in developing safe and intelligent flying robots, and the deployment of large-scale autonomous vehicles raises important questions about efficiency, sustainability, and human-robot interaction. The strategies of companies in the autonomous vehicle space differ in their approaches, and Optimus Ride focuses on using humans to operate vehicles in certain environments, aiming to create efficient, safe, and sustainable transportation systems. The application of autonomous vehicles in transportation-deprived environments has the potential to improve mobility and reclaim space currently used for parking.

Summary & Key Takeaways

• Suresh Carmen, a professor at MIT and co-founder of autonomous vehicle company Optimus Ride, discusses the challenges of autonomous flying versus autonomous driving.

• Autonomous flying is currently easier for consumer drones, but the real benefits of autonomous flying in transportation and logistics pose greater difficulties.

• Large-scale deployments of autonomous vehicles face challenges related to interaction with humans, regulations, business models, and societal issues.