Name: Reactive planning
You will learn: Java/C++, reactive planning
Difficulty: moderate
Excitement: Reactive planning paradigm is often used in Artificial Intelligence and is considered an industry standard for single AIs
Description:

The reactive planning paradigm is often used due to the timely fashion it operates in - the techniques employed range from Finite State Machines (FSMs) to Behavioral Trees and IF-THEN rules. Most models currently used are simplistic to avoid implementation and debugging issues. The basic model of Reactive Planning (RP) is the Hierarchical RP, where IF-THEN plans are organized into a tree-like hierarchy structure.

The goal of this thesis is to implement a newly introduced State Full HRP (SF-HRP) model. The resulting library should be easily incporatible into other projects

There are two distinct themes. Each can be done by one student:

Theme 1 - reserved
C++ implementation - this work is mainly focused on the C++ language. The implementation should be done with optimization and speed in mind, as well as properly applied programming techniques. The implementation should optionally support multi-thread platforms.

Theme 2
Java implementation - this work is mainly focuced on the Java language. The implementation should be done with speed in mind. The resulting code should be well written and easily readable. The implemtation should optionally support remote invoking of the execution (e.g. CORBA etc).

Further Reading: 1. Plch, T.: Action Selection for an Animat. Diploma thesis. MFF, UK, 2009
2. Isla, D.: Handling complexity in Halo 2. In: Gamastura online, 03/11 (2005)
3. Brom, C.: Hierarchical Reactive Planning: Where is its limit? In: Proceedings of MNAS workshop. Edinburgh, Scotland (2005)
4. Bryson, J.: Intelligence by Design: Principles of Modularity and Coordination for Engineering Complex Adaptive Agents. PhD thesis, Massachusetts Institute of Technology (2001)

reactive_planning.txt · Last modified: 2010/10/15 09:14 by jakub_gemrot