growing baby

For months, Mateo has been mastering static equilibrium in assisted standing. We would hold his arms and stand him up (which he would enjoy tremendously, and for a long while), and he’d stand. His legs would be usually straight — potentially locking his knees. Whenever some leg would buckle he would immediately try to straighten them out, and remain in the assisted standing position.

In the last few days, Mateo’s transitioned towards working on the much more complicated dynamic equilibrium problem. In this mode, his legs are usually bent and his butt is wiggly. We are obviously still holding him (or rather he’s holding on to our fingers), but his legs are constantly moving. As the knee bends or as he starts leaning towards one side or another, Mateo will adjust by moving the same leg a bit, or move his butt a little — all to remain standing, but not by locking his joints.

This is a welcome advance, of course. If I was a baby trying to stand up on my own, I would probably figure out next that by moving my feet around I can also affect my equilibrium — instead of keeping them firmly planted, I’d sort of shuffle around. Let’s see if this is what happens.

Dynamic equilibrium is in general much harder than static equilibrium. With static, you just compute the best position of your limbs to cause you to stand. If you miscalculate, you will probably fall. With dynamic equilibrium, you are basically always providing direction to your limbs: a push to the right, a wiggle in the back, a lean to the left, etc, where all of these gradually bring you closer to equilibrium. You may lean over far more than you intended, but that will just cause you compensate more. As long as you can use the wrong direction as input in the corrective command, you’re gaining net equilibrium.

For an example, see this video of a robot (“Big Dog“) maintaining equilibrium while walking on all fours — even as he is kicked, or as he slips.