Episode
Vijay Kumar: Flying Robots
56:57
Published: Sun Sep 08 2019
Description
Vijay Kumar is one of the top roboticists in the world, professor at the University of Pennsylvania, Dean of Penn Engineering, former director of GRASP lab, or the General Robotics, Automation, Sensing and Perception Laboratory at Penn that was established back in 1979, 40 years ago. Vijay is perhaps best known for his work in multi-robot systems (or robot swarms) and micro aerial vehicles, robots that elegantly cooperate in flight under all the uncertainty and challenges that real-world conditions present. This conversation is part of the Artificial Intelligence podcast. If you would like to get more information about this podcast go to https://lexfridman.com/ai or connect with @lexfridman on Twitter, LinkedIn, Facebook, Medium, or YouTube where you can watch the video versions of these conversations. If you enjoy the podcast, please rate it 5 stars on iTunes or support it on Patreon.
Chapters
Vijay Kumar, renowned for his contribution in the field of multi-robot systems and micro aerial vehicles, talks about his experience being a part of a 7,000-pound large hexapod robot powered by hydraulic actuation and controlled by Intel processors.
00:00 - 02:30 (02:30)
Summary
Vijay Kumar, renowned for his contribution in the field of multi-robot systems and micro aerial vehicles, talks about his experience being a part of a 7,000-pound large hexapod robot powered by hydraulic actuation and controlled by Intel processors.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
The ability of small UAVs to maneuver and coordinate with each other in three-dimensional patterns, forming physical shapes in midair, is what makes them beautiful to engineers.
02:30 - 07:29 (04:58)
Summary
The ability of small UAVs to maneuver and coordinate with each other in three-dimensional patterns, forming physical shapes in midair, is what makes them beautiful to engineers. This technology has given rise to the concept of aerial robots.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
The goal is to think about the robots in a low-dimensional space as a cohesive unit, rather than worrying about the individual components, such that the robots can recognize the coordinate system, perform search and rescue operations in dangerous territories, and even count fruit crops in orchards.
07:29 - 19:56 (12:26)
Summary
The goal is to think about the robots in a low-dimensional space as a cohesive unit, rather than worrying about the individual components, such that the robots can recognize the coordinate system, perform search and rescue operations in dangerous territories, and even count fruit crops in orchards.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
The coordination of RPMs is what allows quadcopters to fly, hover, change their orientation, and velocity, and it all comes down to low-cost IMUs.
19:56 - 27:42 (07:45)
Summary
The coordination of RPMs is what allows quadcopters to fly, hover, change their orientation, and velocity, and it all comes down to low-cost IMUs.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
The field of machine learning and reinforcement learning, particularly with the neural network variant of deep reinforcement learning, is making strides in optimizing flight efficiency for flying robots.
27:42 - 32:04 (04:21)
Summary
The field of machine learning and reinforcement learning, particularly with the neural network variant of deep reinforcement learning, is making strides in optimizing flight efficiency for flying robots. While success in flying robots has not completely relied on machine learning, advancements in perception through computer vision and the idea of learning have greatly contributed to flying through constrained spaces.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
The use of data-driven and learning-based approaches in combination with model-based systems is essential to successfully develop autonomous vehicles.
32:04 - 38:43 (06:39)
Summary
The use of data-driven and learning-based approaches in combination with model-based systems is essential to successfully develop autonomous vehicles. However, identifying all corner cases using a single sensing modality and machine learning alone is a tough challenge.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
The speaker discusses the role of collaboration between humans and robots in developing agile flying robots for various purposes, including agriculture and delivery services.
38:43 - 48:01 (09:17)
Summary
The speaker discusses the role of collaboration between humans and robots in developing agile flying robots for various purposes, including agriculture and delivery services. He emphasizes the importance of considering human involvement in the development process.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
Robotics can model humans and assess their state, but it's important to consider ethics and the potential for misuse.
48:01 - 54:28 (06:26)
Summary
Robotics can model humans and assess their state, but it's important to consider ethics and the potential for misuse. Developing the technology in-house can help ensure responsible use.
EpisodeVijay Kumar: Flying Robots
PodcastLex Fridman Podcast
The importance of mathematical foundations and representations in robotics should not be underestimated.
54:28 - 56:59 (02:31)
Summary
The importance of mathematical foundations and representations in robotics should not be underestimated. In order to achieve explainable AI, there must be understandable representations.