Type-2 Fuzzy Logic-Based Linguistic Pursuing Strategy Design and Its Deployment to a Real-World Pursuit Evasion Game

Area of Science:

• Artificial Intelligence
• Robotics
• Game Theory

Background:

• Pursuit-evasion games (PEGs) require dynamic strategies for optimal performance.
• Fuzzy logic offers a framework for decision-making in uncertain environments.
• Type-2 fuzzy logic enhances the handling of uncertainty compared to Type-1.

Purpose of the Study:

• To present a systematic and interpretable design approach for type-2 fuzzy logic-based linguistic pursuing strategies (PSs).
• To enable real-time adjustment of PSs in dynamic PEGs using a type-1 fuzzy decision-making (T1-FDM) mechanism.
• To validate the effectiveness of the developed T2 fuzzy logic-based PSs against human players.

Main Methods:

• Development of a novel type-2 fuzzy logic-based strategy planner (T2-FSP).
• Theoretical investigation of the T2-FSP's input-output mapping to link footprint of uncertainty (FOU) sizes with PSs.
• Design of a T1-FDM mechanism to dynamically tune T2-FSP's FOU sizes for real-time PS adjustment.
• Implementation and validation in a real-world PEG environment.

Main Results:

• Demonstrated that tuning T2 fuzzy set FOU sizes allows design of linguistic PSs, controlling pursuer approach (aggressive/smooth) and side (left/right).
• Established an interpretable relationship between FOU sizes and PSs.
• Showcased real-time adaptability of PSs via the T1-FDM mechanism.
• Achieved satisfactory performance of T2 fuzzy logic-based PSs against human users in comparative experiments.

Conclusions:

• The proposed systematic design approach enables interpretable and tunable type-2 fuzzy logic-based pursuing strategies.
• The T1-FDM mechanism effectively adjusts PSs in real-time for dynamic PEGs.
• The T2 fuzzy logic-based PSs demonstrate robust and satisfactory performance in a real-world pursuit-evasion game context.