Processes that look the same for different reasons.
A portion of my Ph.D. research concerns how to get independently-moving robots to never move too far from one another. There are three basic approaches used to accomplish this in existing work. In a formation, each robot deduces where it needs to be relative to the whole and moves towards that point. In a flock, each robot averages its individual objectives with the need to stay with the group. Switching control laws have robots move however they want most of the time, but switch to moving towards their neighbors when they realize they are getting to far away. I add a fourth approach, using a set of constraints to reduce the options a robot has to those that don’t break the group up and then letting each robot do whatever it likes within that constraint.
When I whip up some basic simulations of these techniques, it can be very hard to tell them apart. In each case, the robots tend to settle into a relatively fixed arrangement and then move as a group. But when I present the different robots with various stressful situations, the differences become clear. The formations act as a single unit; they are unable to navigate forest-like obstacles too large for the group but small enough for the individuals within it. The flocks can be distracted by a single outlying robot, and can be scattered into smaller groups if faced with too many obstacles. The switching laws can cause the entire group to freeze when accosted by moving obstacles, or can fail to switch soon enough and cause the group to scatter. The constraints can compel the robots to preserve the group at the expense of other goals, allowing them to be herded like sheep.
Now, each of these difficulties can be mitigated in various ways; my point is not to discuss the intricacies of keeping robots together. Instead, I want to point out the parallel to every other walk of life, and to people in particular.
During one year of my college career I was set up on over a dozen blind dates. One outcome of that year was this observation: The observation misses a much more significant message one can learn from failed dates, but more on that in a later post. “a blind date teaches you how little your friends actually know you.” They see me moving in my fixed arrangement through the open fields of friendship and they think “that looks like a flocking algorithm. I know just the girl to flock with him!” And then said girl and I meet and she won’t enter the forest and I freeze up in the chaos and it becomes clear to both of us that we’re not running the same algorithm at all.
In the musical “Saturday’s Warrior”I do not recommend the musical. there is a lad who sings “And who am I? Take a look at this kid, a cypher on the wall, not hardly brave at all.” With typical musical irony, the lad turns out to be plenty brave, yet the audience is meant to sympathize with this fellow, and I think on the whole they do. Why? Because most of usI was about to write “all of us” but I don’t actually know that. have no real clue what we ourselves are up to. What algorithm am I using to stay close to my friends? What kind of people do I like to date? How did I decide which shirt to wear this morning? Who am I?
education One of the facts that continually haunts me as I look at educators is that very few people who have learned a subject well have any idea how they learned it, and even less idea how anyone else might learn it. All we can see is the surface, the skin, the evidences of experiences and actions. Beneath it lies the unseen truth, vastly more complicated and nearly unknowable.
Who am I? I’m the guy who just typed this web log post. How did I do so, and why? Not even I actually know that.