Musicians and Robots

In Alpin Hong’s TEDx video on YouTube, he plays the opening of Mozart’sTwinkle Little Star Variation.” (“Twinkle Twinkle Death Star” starts at 14:15) It’s a fast, intricate piece which requires sustained, robotic precision. There are many styles besides classical which sometimes require robotic technical precision. But classical music is often a lot more technically demanding than popular styles.

How we as humans compare with robots depends on what kind of robot you’re talking about. There are two trends today in the development of robots. Industrial robots, like Motoman robotic arms, are based on precise, pre-programmed motion control. There are robots that can play musical instruments, but they can’t react to any changes in the positioning of the instrument. Everything has to be carefully positioned and calibrated. All they can do is perform exact movements very reliably. ASIMO is a humanoid robot that was developed by Honda a few years back. ASIMO is capable of complex pre programmed dance moves, but it can’t play soccer with other robots.

The other trend is toward robots that can react to changes in their environment, besides just shutting down when someone gets too close. Baxter is a robot from Rethink Robotics. Baxter is capable of making decisions based on what it learns. Also, Baxter has an acute kinesthetic sense, so it can tell when a bolt is about to strip from being turned too tight, just like us. Baxter’s movements are based on vision and memory, just like us. It is continuously monitoring and correcting based on what it sees and feels, just like us. However, Baxter’s movements are slower, and more gentle than ASIMO’s, or a Motoman.

Every concert pianist strives to become like ASIMO, capable of performing moves exactly the same way each time they perform them. But ASIMO can’t correct for mistakes. By our nature, humans are more like Baxters. We have to rely on our vision, and our senses to deal with unanticipated conditions in our environment. We have to monitor and correct continuously. There are limits to how much humans can emulate robots.

As much as we would like to emulate ASIMO’s robotic precision, because we’re human, that level of sustained precision isn’t possible. As much as we may try to be ASIMOs, we’re really all Baxters.

Alpin Hong says that the key to recovering from mistakes is to be ready to improvise in the moment, based on a framework ingrained in your memory. Because we’re human we need a strategy for dealing with mistakes that isn’t based on just more robotic precision. Sometimes as a last resort you have to improvise your way back based on concepts.

Humans aren’t that good at performing long chains of complex sequences, unless we learn by breaking those sequences into smaller parts. It’s important, when you’re learning new music, to mentally break it down into sections. Musicians learning a new piece often say, “I can’t pick up in the middle, I have to start at the beginning and play all the way through.” That kind of mental inflexibility is dangerous. It’s good to mentally break music into small sections, and then put the sections back together. The more ways you can break a piece up mentally, the better you know the piece, and the less likely you are to get thrown off by a momentary lapse in memory.

© 2019, 2020 Greg Varhaug