Return to RESEARCH PAGE
Intro to our Robots & our Robotics Curriculum
I. Online Introduction to Robotics
II. Designing Robots
III. Knight Rover Robotics
Artificial Intelligence programs can be very effective in helping to think about the nature of minds, both natural and artificial. But mere programs, lacking the ability to manipulate objects in the world, lacking the ability to 'do' anything, will prove to be an inadequate model for human cognition. Thus our interest in machines that engage intelligently with the world.
I. Online Introduction to Robotics
This website has many modules to introduce students to a wide range of topics in AI and robotics. All of the modules on topics related to artificial intelligence and robotics can be found in our complete "Curriculum" sorted by "Category": The relevant categories are the first ("AI, Robotics and Computational Models") and the third ("Foundational Issues") categories on this list:
Curriculum (by Categories)
In addition to these separate modules, we also have integrated many of the modules, added new content, and created one large "Introduction to Robotics" module:
 |
Introduction to Robotics (with Virtual Labs)This "Introduction to Robotics" has integrated many of The Mind Project's modules on AI and robotics into a general introduction that focuses attention on a broad range of different robot-types and shows how they are currently being used in the medical field. Three separate virtual robotics labs challenge the use to build a robot -- one is a classic "top-down" architecture, two are "bottom-up" or behavior based. as well as many animations and videos are included. |
II. Designing Robots: Top-Down vs. Bottom Up
One good place to start when building an intelligent machine is to consider two influential (and fundamentally quite different) models for robotic design.
A. Top-Down Robotics
Using the first strategy (top-down), you could write a sophisticated computer program that 'builds in' all of the desired cognitive abilities in the form of computational states that explicitly model mental states. For example, you might design a system with (a) sensory devices that take in information from the environment and support, (b) a symbolic representation of the enviornment which is used by (c) the cognitive "agent" that generates "beliefs" about the world and behaves in response to those beliefs together with its motivating "desires." The behavior of this machine is determined by its "mental states" -- its "beliefs" and its "desires" (or at least computational states which have a systactic structure analogous to human mental states.)
One of the first immersive experiences that we created is our "Virtual (Top-Down) Robotics Lab" where the user builds a mobile robot and programs it to recycle Coke bottles.
Virtual Robotics Lab ("Top-Down")
Users build and program a mobile robot, assembling all components, constructing a robotic arm, writing scripts to direct the arm to pick up a Coke bottle, writing scripts to steer the robot's wheels, loading "beliefs" into the main AI engine (ProtoThinker), and watching the Iris.4 robot move through the lab, recycle the Coke bottle (because in its "language of thought," it is a committed environmentalist).
The virtual robot in the lab above is based on one of our physical "Iris" robots. From 1995 thru 2014 undergraduates at Illinois State University have built 4 versions of "IRIS," each one adding capacities that exceeded the previous version. At the heart of the IRIS robots is the ProtoThinker software. ProtoThinker (or "PT" for short) is an artificial agent written in Prolog, that learns (from the people it talks to), acquires beliefs and desires about the world, and responds in language and in behavior (with the addition of end-effectors like motorized wheels and robotic arms). A history of the IRIS series robots (now retired) is found here:
IRIS Robots (Version 1-4)
The first IRIS robot (IRIS 1.0) was created in 1995 and consisted of the ProtoThinker software, an open-ended set of beliefs and desires (that grew as it engaged with the world) and one robotic arm. You can meet IRIS in this short video: The name where you want to jump
Meet IRIS 1.0
IRIS is possible because of the genius of John Barker's ProtoThinker software.
B. Bottom-Up (or "behavior-based") Robotics
The second strategy (bottom-up), requires nothing so computationally complex as the AI software that typically controls top-down robots. Instead this type of machine is equipped with many simple (even 'stupid') stimulus-response mechanisms. If these "behaviors" are structured into a productive hierarchy, then these simple devices will perform their function at the proper time, in response to appropriate environmental conditions, resulting in behavior that is remarkably complex, even 'intelligent' -- and yet without anything that looks remotely like "mental states" as we normally understand them.
Ken Livingston and Josh de Leeuw at Vassar College created our bottom-up or behavior-based virtual robotics lab. This is a wonderfully designed lab that is easy to use and gives a good introduction to behvavior-based robotics in a very short period of time.
Virtual Behavior-based (or "Bottom-up") Robotics Lab
Behavior-based (or "bottom-up") robotic design does not equip the robot with anything like a "mind" (with beliefs and desires) but instead uses a relatively small number of stimulus-response mechanisms, each by itself just a stupid behavior. But if you can create the right "hierarchy" -- giving each behavior the right rank-ordered priority, you can get intelligent behavior. Observe, design & test your own robots.
To complement both the bottom-up and the top-down robotics labs, we have launched a new physical robot project: The Knight Rover Initiative -- which is described below.
III. Knight Rover Robotics
When you want to learn about robotics where do you begin?
In the Spring of 2017, we launched a brand new, long-term, project. The Knight Rover robotic system is designed for use in the classroom, making use of modestly priced readily available robot kits and our new ROOK software. In the Spring of 2018, we have a team of 10 students (9 college-age, 1 high school) working to create activities for use in middle school, high school, and college classrooms.
ROOK has been created for use in educational settings and allows students to engage with the software regardless of their level of expertise. At the most basic level, students learn to use the software to build a robot with certain capabilities. The deeper they dig the more they learn about computer programming and the greater freedom they achieve to re-write and expand the computational systems controlling their machine.
One long term goal is to create an inventory of cognitive systems that include some that are top-down and some that are bottom-up. As students struggle with the question: "Is human intelligence primarily top-down, bottom-up, or some complex integration of the two?" -- they will be able to design robots that include both types of cognitive systems.
You will find the current development page and much more about the "Knight Rover" robotics team on its homepage:
Knight Rover Robotics
Let us know if you have your own robotics group with work you would like to share with us.