SE367: Introduction to Cognitive Science

In this homework, you will consider the task of picking up a pencil. Here is a detailed view, from some students in this course 18 months back, on how to instruct a "robot" to execute this task:

http://www.cse.iitk.ac.in/users/se367/11/se367/hgini/hw2/

The set of instructions include statements such as:

 
    Identify coordinates of pencil in XY plane
    Hold pencil at p = 2cm , with M and T affectors
    Rotate pencil (about Z-axis, passing through p = 0), till (x=0,y=0)
        coordinate lies on P-axis
    Move pencil along Z-axis by 20cm
        Source point- = defined as centre of trough between I and T
        affectors 
    Rotate point of contact between I-affector and object, about axis
        defined by line passing through points of contact of M and T
        affectors (with object), by an angle such that P-axis lies at
        perpendicular distance of 'r' from 'source point'.
    [...]

Now consider the state-of-the-art in having robots do such tasks The 2nd instruction "hold pencil at XXX" is itself barely manageable, as you can see in the video of a robot learning to pick up a pen (by Mrinal Kalakrishnan). Of course, this robot has no sensors on its fingers, and the main focus of this work is to use reinforcement learning to optimize a grasping strategy that emphasizes smooth motions, and is rewarded whenever the fingers that do not cross a threshold (pen has not slipped). Being able to hold on to it longer gives a better reward, which provides a gradient for the learning algorithm.

Reports should be individual, but based on discussions in the group. Discussions should be held before Wednesday Feb 13 midnight. The reports are due by Friday Feb 15.

Also watch at least these two parts 2. memory and 3. process, from the video My Brilliant Brain on chess player Susan Polgar.

I will encourage you to watch the video in a group (need not be only from this course) -- it raises many important questions regarding automaticity and expertise.

QUESTIONS to answer in your reports:

1. Which two instructions in the "programming language" of the 2011 HW would be the most difficult for robots to follow?
2. The robot following the learning paradigm as in Kalakrishnan is clearly gaining some expertise. Which aspects of the execution may be called implicit or automatic, and which aspects may be more explicit? What could be the "chunks" in this structure?
3. Comment on whether human learning may also be following similar "reward" based processes? Consider the learning process for the fire-fighting expert who knows how to fight complex fires.