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

Transcript

Read and summarize the transcript of this video on Glasp Reader (beta).

Summary

This conversation is with Russ Tedrick, a roboticist and professor at MIT and vice president of robotics research at Toyota Research Institute. They discuss topics such as the most beautiful motion of animals and robots, the concept of passive dynamic walkers, the importance of rigorous thinking in robotics and the role of deep learning in the field.

Questions & Answers

Q: What is the most beautiful motion of an animal or robot that you've ever seen?

I think the most beautiful motion of a robot has to be the passive dynamic walkers. The ones built by Steve Collins and Andy Ruina at Cornell, in particular, were remarkable. These 3D walking machines were powered only by gravity and had a natural, human-like motion that was graceful and elegant.

Q: How does a passive dynamic walker work?

The simplest model for a passive dynamic walker is similar to a slinky, but with elaborate spokes. When given a push, the walker falls down a ramp and each time its foot hits the ground, it loses a little energy. However, with each step forward, it gains a little energy. These two energy losses and gains can come into perfect balance, resulting in a stable periodic motion.

Q: How do you design a passive dynamic walker?

Designing a passive dynamic walker requires a combination of science and art. There is a science to getting close to the solution using models and analysis. However, there are also artistic elements, such as tuning the suction cups at the knees to stick and release at the right times. It is a delicate balance between mechanics and control that ultimately creates a beautiful and efficient robot.

Q: Are animals more driven by dynamics or control?

Animals, including humans, are a combination of dynamics and control. While control is important for adapting to different situations, dynamics play a crucial role in efficient and natural movement. Animals have evolved to optimize their mechanics and adapt to their environments, making them efficient and capable of complex movements.

Q: Why do humans walk on two feet?

Walking on two feet, or bipedal locomotion, is advantageous for humans in several ways. One theory suggests that it allowed early humans to carry food back to their tribes, promoting cooperation and survival. Additionally, bipedal locomotion makes it easier to navigate various terrains, as the world is built for the human form.

Q: Is there a practical advantage to building robots that walk on two feet?

Building robots that walk on two feet, or bipeds, can have practical advantages in certain situations. The human form allows robots to navigate environments designed for humans, making them more versatile. However, as we continue to develop robots and shape our environments, other form factors may become more suitable for specific tasks.

Q: Why do you run barefoot?

I started running barefoot because of my work in studying walking robots. It helped me develop a better understanding of human and robot movement. Running barefoot provides immediate feedback on my running form and allows me to land softly, protecting my joints. It also allows me to connect with nature and decompress from work.

Q: What tips do you have for someone interested in trying barefoot running?

Start slow and be patient. Take the time to build up the strength and conditioning of your feet and lower legs. Consider starting with minimal shoes or running sandals to provide some protection while still allowing a more natural running gait. Listen to your body and pay attention to any pain or discomfort, adjusting your running form as needed.

Q: What is the most beautiful idea in robotics?

The most beautiful idea in robotics, for me, is the clarity of thought that comes from rigorous thinking, particularly in control theory. The ability to apply mathematical principles to complex and messy problems is awe-inspiring. While deep learning has its merits, there is something to be said about the power of clear and rigorous thinking in robotics.

Q: How can we encourage rigorous thinking in robotics and related fields?

One way to encourage rigorous thinking in robotics is through education. Taking courses in analysis and other foundational topics can help develop critical thinking and analytical skills. It is also important to value and promote rigorous thinking in research and academic settings, ensuring that it is recognized and rewarded.

Q: Can deep learning replace rigorous thinking in robotics and related fields?

Deep learning has undoubtedly revolutionized the field of robotics and has made it easier to achieve certain tasks. However, there are limitations to deep learning, especially when it comes to understanding the underlying principles and limitations of a system. There is still a need for rigorous thinking and understanding of the fundamentals to push the boundaries of robotics further.

Takeaways

Rigorous thinking is a beautiful aspect of robotics that should not be overshadowed by the ease and allure of deep learning. While deep learning has its merits, clear and rigorous thinking is crucial in understanding the fundamentals and limitations of a system. Education and recognition of rigorous thinking are essential in advancing the field of robotics. Additionally, exploring alternate forms of locomotion, such as barefoot running, can provide valuable insights into human and robot movement.