- Teach students to analyze navigation and control tasks
- Teach students to break complex tasks into a set of small scenarios
- Teach students to express task requirements as a set of rules
- Teach students to express task solutions as a set of scenarios where each scenario represents a flow of rule-based service invocations
- Teach students to express rules in a logically correct way using OpenCyc and specialized graphical interface that describes robot sensors and manipulators as well as navigation options.
The system consists of:
1. A java-based software application with integrated OpenCyc knowledgebase, graphical user interface, scenario player, and a communicator, that transfers (robot-related) instructions to mobile robots over wireless communications.
2. A set of mechanical robots with a set of sensors, manipulators, and wireless interface capable to accept instructions from the host system.
One of the popular models in Artificial Intelligence study is a Wumpus World.
The full definition of this model is given in the most popular AI text .
Here may be found an exact definition of the Environment which consists of squares and different possible perceptions on each square according to Wumpus, Pit or Gold position Breeze, Glitter or Smell. The goal of an intelligent agent is to get gold back to start without entering Pit or Wumpus square.
The agent actions are Turns Left and Right, Shoot, Grab and Release gold.
Fig.1. The Wumpus World
The Wumpus World is used to study Action and Reasoning while building some kind of Knowledge Representation. A lot of software simulations were implemented as projects during AI courses.
Our purpose is developing a "real" simulation of the Wumpus World. The Environment will consist of rectangular plates located on the floor with Autonomous Mobile robot moving across and collecting information to build a Knowledge Base.
The mobile agent will be build using a new generation of the Lego Mindstorms NXT (accessible from August this year).
Fig. 2. LEGO MINDSTORMS NXT includes (announced by Lego Ltd.):
- All-new NXT intelligent brick
- Bluetooth and USB connection
- 3 interactive servo motors feature inbuilt rotation sensors
- New ultrasonic sensor makes robots see by responding to movement
- New sound sensor enables robots to react to sound commands, including sound pattern and tone recognition
- Improved light sensor detects different colors and light intensity
- Improved touch sensor reacts to touch or release and allows robots to feel
- 519 hand-selected, stylized elements from the LEGO TECHNIC building system ensure robot creations will be sturdy and durable while also looking authentic
A mobile agent (One of Mindstorms NXT robots, assembled and programmed by students in Afeka college) uses a simple embedded program that analyses sensor inputs and sends a message to a host system via Bluetooth communications. The host system (based on infrastructure developed by ITS) uses OpenCyc Knowledge Base and performs inference process to find a Wumpus position.
The correspondent instruction is sent back to the robot to perform a proper action.
The host system is designed by ITS, Inc., http://javaschool.com.
The host system uses OpenCyc as its business intelligence engine that drives the application ,.
The main components of the host system are displayed on the FIG. 3. below.
Fig. 3. The host system by Internet Technology University, ITU
There are main components: Knowledge Service Component, Service Connector, Scenario Player, and Presenter.
The Knowledge Service Component consists of the OpenCyc and OpenCyc Connector that extends existing OpenCyc Java API.
The Scenario Player interprets XML-based business scenarios and sends instructions to robots via the Communicator
(a part of the Presenter component).
Scenarios can include references to rules and conditions.
In this case ScenarioPlayer interacts with the OpenCyc (via the OpenCyc Connector).
The Performer displays a set of robot sensor and manipulation options, and allows us to build a flow of conditional instructions.
The Formatter translates conditional instructions in the XML-based scenarios and also helps students creating CycL assertions
that express task requirements and capture complex solutions.
We believe that the project is absolutely doable in frame of above definition.
It may be further extended to Multiagent System where few robots are performing a common task to find a Wumpus while sharing the Knowledge Base and communicating by Net, Bluetooth or Cellular phone connection.
An important part of the project is a Learning Subsystem.
Under and Post graduate university students will learn an application of Modern Logic with OpenSyc to such fields as
- Introduction to Artificial Intelligence user friendly environment to build and use a Knowledge Base and activation of built-in Inference algorithms
- Robotics different navigational algorithms for autonomous robots.
- Multi-agent Systems experiments with different behavioral models of mobile agents.
The additional value of the project is in its integrated nature leading Knowledge Base technology, attractive field of Artificial Intelligence study, modern worldwide recognized robotics kits.