Design Perceptive

—/////—

From Scientific American magazine
5 Future Robotics Expeditions and What They Could Reveal.
This is a slight misnomer.  The only future robot is the Mars Science Laboratory.  The remaining have been already launched.  Yet, they are still on their respective journeys to their destinations.

—/////—

This was my Tuesday-morning Ted Talk watching experience this week.  It resonates with my current interests in Computation Creativity.  It’s a brief lecture by Tim Brown running about 28 minutes long.  He talks about play in the form of exploration, building and role-playing and their roles in creativity.  You should take a gander when you have the time.

—/////—

We here at Design Perceptive are always interested in the way humans perceive things.  It gives us direction and clues in designing towards artificial perceptive agents.  So, the interest here is to ask how and why do perceptual illusions work?  Consider the illusion in this article – The importance of perceptual illusion research from Cognitive Daily.

—/////—

The brain is amazing.   I really think many people (members of the medical/scientific community excepted) misunderstand how amazingly robust the brain really is.  This article in this Neurophilosophy blog shows how the brain can just keep going and going and going.

—/////—

I love the Study Hacks blog.  This article address the age old problem of procrastination and analysis paralysis.  I’ve run across this problem.  You most likely have as well.  Almost all of the classes in my major have semester projects, but this Ice-bath method could be applied to individual homework assignments as well.

I like the concept of a minimalism when it comes to the home. It appeals to the senses by simplicity. And, in my opinion, one characteristic of a minimalist home is that it requires minimal effort to run and maintain. In other words, it is very hands off. So, I’ve been working with intelligent control systems, adding them as I can. Recently, I installed a simple timer-based irrigation system for my garden and patio.

Now, I’ve started looking at controls for the HVAC. I want to be comfortable, but I don’t want to have to deal with the thermostat anymore. I also want to save money, but I don’t want the hassle of trying to figure out everyone’s schedules and programming the system for that. Now, these goals are somewhat contradictory. If I wanted to maximize my money savings, I simply turn off the AC. But, that costs me in terms of comfort. If I want to be comfortable I run the system, and I am going to incur operating expense. There’s a balance here somewhere.

So, after a little foray into a Research Experience for Undergrads, where I did some research into intelligent control systems, I’ve decided to start wiring up my house. I have a good thermostat that gives me the ability to control the HVAC via my computer. I can also read information about the temperature in the house.

I also need to keep track of who is home and who isn’t. To do this, I’ve enlisted the help of Google and my cell phone. Every so often, my phone tells Google where it is. I can pull this information straight off the web through a nifty URL Google provides. I currently have my home computer doing this every 10 minutes. I’m going to let it run tonight and all day tomorrow.

I propose to use this mechanism to train an artificial neural network designed to learn my occupancy patterns. This can be used to determine if I’m going to be home at any particular time, giving my HVAC something to work with to determine when it should operate.

The use of potential fields in path planning is well established in the field of artificial intelligence.  In general, a physical space is modelled by a field of vectors where goals are attractors and obstacles are repellers.  Determining the direction to move for a mobile robot is then a matter of following the vector.

As intelligent agents, humans are constantly being pulled about by contending forces.  Our own beliefs and intentions betray us when acting as motivators to our own behavior.  I may horribly detest flying to the point of aversion, but the desire to flee conflict or danger may override that aversion and cause me to get on an airplane.  We tend to manage these conflict in a number of ways, including prioritization of goals.  Where one goal might be a low priority now, might become high priority later by virtue of time or other factors.

I am proposing the use of potential fields as a model for complex motivational system.  Under this model, behavioral states are modeled as attractors and repellers much like goals and obstacles.  Rather than modelling a physical space defined by the x,y,z dimensions, I propose a  space that defines relevancy.  In short, potential fields provide a means for modeling relevancy.  It also provides a means of dealing with conflicting goals.

I’m in the second, and final, semester of Senior Design in the Computer Science program at the University of Texas at Arlington.  One of the requirements of this curriculum is to sit a panel discussion.  Each member of the panel must give a 10 – 15 minute presentation and then be prepared to discuss and answer questions.  My panel sits about 4 weeks from now.  Naturally, but unlike some of my fellow students, I’m preparing now.  I mean, this type of thing is part of what I want to do for a career.

Anyways, the topic I’ve been assigned is “What are the social implications of artificial general intelligence?”  Yeah, I know I could fill an entire semester with this topic, but I only have an hour in which I must present and represent.  So, I’m trying to come up with a decent presentation that introduces the concept of AGI, discusses some of the popular conceptions of what happens when we do create AGI and what the social implications might be.

Going through this exercise is synthesizing some of my own thoughts.  Though they might not be novel, I’m writing them down here as a way to remember the path; a breadcrumb if you will.  It may be that I find support in the literature for these ideas, find that someone has already thought of these, etc.  That would be fine – I just need to get them out.

First, it seems to me that motivation is a key element in AGI.  Maslow’s Hierarchy of Needs might then play an interesting role as part of a motivational system for a generally intelligent agent.   I’ll be looking into this concept and discussing more of it here later.

I spent Saturday evening on a party barge on Lake Grapevine.  My wife and I had been invited by some friends from the Reality Abatement Bureau to go and listen to some dub step played by some of the coolest DJ’s in the metroplex.  Much to my surprise, my good friend David Hanson was there as a last minute step-in for someone else who decided not to come.

It’s always a good time when I get to speak with Dr. Hanson.  We spoke of Genetic Algorithms and how evolutionary development (ontogenesis), while continuous, is punctuated.  We particularly spoke of evolution at the social level and using a MMORPG as a breeding ground for social agents where player characters (controlled by humans, of course) provided the fitness score for these agents.