Summary of Russ Tedrake: Underactuated Robotics, Control, Dynamics and Touch | Lex Fridman Podcast #114

This is an AI generated summary. There may be inaccuracies.
Summarize another video · Purchase summarize.tech Premium

00:00:00 - 01:00:00

In this video, Russ Tedrake discusses the challenges involved in developing underactuated robotics, control, dynamics, and touch. He also discusses the team's solution of fastqp, a lightweight, fast layer that would stabilize the robot in the event of a slowdown.

  • 00:00:00 Russ Tedrake discusses the beauty of underactuated robotics, discussing the passive dynamic walkers created by Steve Collins and Andy Rowena at Cornell. He also discusses the various countries with their own versions of the Office, which all use ExpressVPN to control where the site thinks he is located.
  • 00:05:00 Russ Tedrake discusses how underactuated robotics, control, dynamics, and touch can create a more human-like movement. He also mentions Honda's ASIMO as an example of how successful this can be.
  • 00:10:00 George Lotter and Mike Tranifilo studied fish swimming in a water tunnel and found that if a fish is behind a rock, its gate changes and it starts swimming upstream. They also did an experiment where they tied a dead fish to a rock and watched it swim. The dead fish started swimming upstream due to its mechanics resonating with the vortex street. The video is worth watching for the clincher experiment where the dead fish starts swimming upstream even though it has no brain and no motors.
  • 00:15:00 Russ Tedrake discusses how the mechanics of our bodies have evolved over time, and how exercise can help improve them. He talks about how his fifth metacarpal was lost in a car accident many years ago, and how it has changed the geometry of his bone to adapt to the new circumstances. He also discusses some of the evolutionary arguments for walking on two feet, and how they can be flawed. He recommends a book by Nicholas Taleb about the concept of "fooled by randomness."
  • 00:20:00 Russ Tedrake discusses his work on bipedal walking and underactuated robotics. He notes that the main reason he spent so much time on the topic was because it was difficult and it challenged control theory. He also notes that running barefoot has taught him a lot about human and robot movement, specifically how to properly take small steps and land softly.
  • 00:25:00 Russ Tedrake discusses his experience running barefoot and biking to work, and how it has helped him to lead a more balanced life. He also recommends reading Ken Bob's book, Barefoot Running, to learn more about the benefits of running barefoot.
  • 00:30:00 Russ Tedrake discusses how his time spent studying control theory has helped him understand complex problems more clearly. He also discusses how deep learning is an example of clear thinking in action.
  • 00:35:00 Russ Tedrake discusses the importance of rigorous thinking in various fields, including robotics and machine learning. He also offers advice on how to practice rigorous thinking.
  • 00:40:00 Russ Tedrake discusses the scope and limitations of underactuated robotics, control, dynamics, and touch. He argues that learning is the route to achieving a simpler, more intuitive understanding of complex systems.
  • 00:45:00 Russ Tedrake, a professor of robotics at Boston University, discusses the challenges of making humanoid robots that are people-like, durable, and autonomous. The team that won the right to work on one of the Boston Dynamics robots in the final challenge was composed of MIT students.
  • 00:50:00 In this video, Russ Tedrake describes the challenges involved in developing underactuated robotics, control, dynamics, and touch. He also discusses the team's solution of fastqp, a lightweight, fast layer that would stabilize the robot in the event of a slowdown.
  • 00:55:00 Russ Tedrake discusses how robotics people are willing to push themselves to the limit in order to be happy. He says this is what makes them happy and successful, and it is something that is fundamental to his thinking. He talks about the team that won the DARPA challenge and how they were able to do incredible engineering and testing.

01:00:00 - 02:00:00

Russ Tedrake discusses the challenges of underactuated robotics, control and dynamics, and how deep learning can be used to improve these aspects. He also mentions how humans are still able to manipulate contact with the world to their advantage.

  • 01:00:00 Russ Tedrake explains how the difficulty of controlling a large, heavy robot in close quarters led to some of the robot's falls during the DARPA Robotics Challenge. He also discusses the importance of pre-launch checklists and how his team relied on motion planning to get the robot up and over obstacles.
  • 01:05:00 Russ Tedrake discusses the challenges of underactuated robotics, control and dynamics, and how deep learning can be used to improve these aspects. He also mentions how humans are still able to manipulate contact with the world to their advantage.
  • 01:10:00 Russ Tedrake discusses how underactuated robotics can lead to difficulties with decision-making and control, as well as the fear of injury and the motivation to excel.
  • 01:15:00 Russ Tedrake talks about how the goal of the 2050 deadline for human players to be beaten by robots is still the goal, and that the focus for robotics should be on mixed martial arts and game playing. He also talks about the Black Mirror episode and how it has changed the way people watch robots.
  • 01:20:00 Russ Tedrake discusses the fear of robots, and how it may be due to our cultural experiences and the stories we tell ourselves. He is a technological optimist and believes that despite the possible risks, technology will continue to evolve and improve.
  • 01:25:00 Russ Tedrake discusses underactuated robotics, control, dynamics and touch in relation to anthropomorphism and the fear of artificial intelligence becoming too intelligent. He argues that, while there are many scary aspects to our future with AI, we should not be scared of the robot, as it will be a part of us.Boston Dynamics is a leading company in the development of underactuated robotics and its work is inspiring the world.
  • 01:30:00 Russ Tedrake discusses the difficulties of making money with robotics companies, and how the success of Boston Dynamics demonstrates that there is still a lot of good work being done in the field. He talks about how anthropomorphization can add another layer of human connection and how robots can be controlled.
  • 01:35:00 Russ Tedrake discusses the successes of robotics, discussing how dynamics and touch are important in creating successful robots. He goes on to discuss how contact is difficult to simulate and how this limits the capabilities of robots.
  • 01:40:00 Russ Tedrake discusses the complexities of contact dynamics and touch, explaining that many contact configurations result in different control systems. He discusses the difficulties of being able to account for all possible contact configurations in controllers and discusses the potential for developing more robust controls through considering every possible contact configuration.
  • 01:45:00 Russ Tedrake discusses the challenges of controlling robots with dexterous hands, pointing out that even if every possible order the hand came into contact with the object was enumerated, the robot would still be at a disadvantage because it cannot simulate contact accurately in simulators. He then introduces Drake, a simulator that allows users to write controllers in a high-level language, as well as system modeling language that helps in understanding the inner workings of complex systems.
  • 01:50:00 Russ Tedrake discusses the challenges of simulating physical interactions between robots and objects in a controlled and reliable way. Tri Robotics Institute has made significant improvements to drake, a simulation engine used to study these interactions.
  • 01:55:00 Russ Tedrake discusses the importance of control theory in the field of robotics, and how rare event simulation can be used to help improve performance. He also mentions the potential for fleets of robots to learn together, and how humans may enjoy this type of learning.

02:00:00 - 02:45:00

Russ Tedrake discusses underactuated robotics, which is the view of the world as an underactuated system. He stresses the importance of details and state representation in robot control and how manipulation tasks can be done without requiring full actuation. He also discusses the importance of touch in robotics and how it can improve the interaction between humans and machines.

  • 02:00:00 Russ Tedrake discusses the various projects at Toyota Research Institute (TRI), including their work on home robotics, advanced materials discovery, and aging in place. He believes that there is a lot of potential for robotics in the home, and thatSoft Robotics is a key technology for advancing this trend.
  • 02:05:00 Underactuated robotics refers to a type of robotics where a mechanical system has fewer degrees of freedom than an actuator-driven robot. In this way, underactuated robots are more like humans, in that they are not completely under the control of their actuators. This lecture discusses the importance of control in robotics, and how underactuated robots present a challenge in achieving it.
  • 02:10:00 Russ Tedrake discusses the challenges of controlling an underactuated system, and how deep learning has made this possible.
  • 02:15:00 Russ Tedrake discusses the idea of underactuated robotics, which is the view of the world as an underactuated system. He stresses the importance of details and state representation in robot control and how manipulation tasks can be done without requiring full actuation.
  • 02:20:00 Russ Tedrake discusses the importance of touch in robotics and how it can improve the interaction between humans and machines. He also discusses how humans can anthropomorphize machines, and how we need to be careful about safety when building robots.
  • 02:25:00 Russ Tedrake discusses his research on underactuated robotics, control, dynamics, and touch. He recommends reading "Ai Superpowers" and "Sapiens, Sapiens: A Brief History of Humankind".
  • 02:30:00 Russ Tedrake discusses the idea that books and papers can be "dialogue between you and the author," and how this can be a valuable way to learn. He also discusses the current trend of online lectures, and how they can be useful for students.
  • 02:35:00 Russ Tedrake discusses the potential benefits of teaching underactuated robotics in a way that is both patient and concise. He also references Joe Rogan, a comedian who has been inspirational to him in terms of conversation. He stresses the importance of variety in a conversation, mentioning that comedians can be a great source of inspiration for someone looking to improve their storytelling skills.
  • 02:40:00 Russ Tedrake discusses underactuated robotics, control, dynamics, and touch in relation to life and robotics. He recommends that young people learn math and engineering in order to be able to think deeply and critically. He also discusses his experiences at MIT, where he became interested in computers and began his current career.
  • 02:45:00 Russ Tedrake discusses underactuated robotics, control, dynamics, and touch in this Lex Fridman podcast. He discusses how MIT's culture demanded more of him mathematically and in critical thinking, and how he never considered quitting despite struggling at times. He talks about how robots are important for space exploration, and how humans should focus on sending more robots instead of relying on humans in space.

Copyright © 2025 Summarize, LLC. All rights reserved. · Terms of Service · Privacy Policy · As an Amazon Associate, summarize.tech earns from qualifying purchases.