While trying to think about tai chi in a scientific context, I came up with the idea of a robotics thought experiment framework. In other words, what would it mean for robots to do tai chi? In order to train a robot (which is essentially a computer) to do tai chi, hippy style energies waffle wouldn't work. Instead you would have to use precise algorithms, definitions and terminologies and have to explain tai chi in the language of science.
It is often said that you only really understand the limitations of your own knowledge when you try to teach someone else. By extension therefore if we take the ultimate "dumb" person, a robot, and metaphorically try to teach it tai chi, we are in fact deepening our own understanding. As my old favourite maths teacher used to say:
It is often said that you only really understand the limitations of your own knowledge when you try to teach someone else. By extension therefore if we take the ultimate "dumb" person, a robot, and metaphorically try to teach it tai chi, we are in fact deepening our own understanding. As my old favourite maths teacher used to say:
"If the answer is obvious, then it is either trivial to explain, in which case you should do so; or else it is non-trivial, in which case the explanation is essential."Now there are a number of ways to approach this problem, and I have come up with four which I sometime refer to in my posts:
- Benchmarking: Tai chi (or the other internal arts) would perform an ideal benchmarking test case for the mechanical abilities of a robot. The whole body coordination and complexities of balance ensure that to perform a tai chi form properly would be exceptionally challenging. This can be seen as a direct analogy to the Turing test, but for robotics. If you could do tai chi push hands with a robot and it felt to you like doing it with a human, to all intents and purposes the robot is capable of doing tai chi.
- Minimal System: Do the principles of the internal arts have any meaning outside of the context of a human body? Can internal arts principles be applied to non-humans (e.g. animals)? Taking this a bit further, given that we can construct robots that are humanoid in form, can they use internal arts principles? What are the minimal degrees of freedom that are required of these robots to be able to exploit or perform the internal arts? It should be noted that this context is different to just pre-programming a sequence of movements (that does not require any thought or reactivity), but rather the dynamic utilisation of the principles as behaviours or reflexes. In CGI simulation terms this is already possible as the troublesome robotics hardware can be bypassed. CGI agents (i.e. simulated robots) can already be endowed with a set of behaviours which result in realistic movements and actions, for example natural motion. I don't see any reason why they couldn't do tai chi.
- Emergent Concentration: What does it mean to meditate or concentrate in robotics? After all, in many ways a computer could be considered an ideal meditator as it is able to "empty its mind" and just revert to something akin to a system idle process. More generally, however, this point is about emergence. Although we commonly think of ourselves as having a single mind, this is in fact an illusion (the dissolution of ego in Buddhist terms). A strong theory of mind is the that humans brains are emergent systems with hundreds of thoughts and ideas competing for attention all the time in the subconscious. What emerges from this soup is the strongest signal, which we call conscious. Robotics embraces this philosophy as it is impossible to have some central controller coordinating every aspect of a complex robot. What is needed instead is a distributed and emergent control structure. How well this emergent system can be controlled is good analogy to humans ability to concentrate or "be at peace".
- Robot Chi: If we consider the thought experiment whereby we have a robot that is a perfect anatomical match for a human (in terms of joint and muscle structure), what are the software algorithms it needs in order for it to be able to perform tai chi? This raises an interesting question: would it be possible to program this robot to perform the internal arts from a purely biomechanical perspective? If the answer to this is yes, then this would prove that there is no such thing as chi, as it is not a necessary component of the system. If however it was impossible to program the humanoid robot using only biomechanics, then this by contrast would prove the existence of chi, as an actual tangible property rather than as an abstract visualisation/model of the world. Perhaps chi is actually an analogy (and opposite) to pain? When you experience pain, it is the body's way of telling you that things are incorrectly aligned/not performing their correct function. When you experience chi, it is the body's way of telling you that things are performing optimally. The more chi you experience, the more optimally your body is functioning.
