Abstract:
Autonomous mobile robot navigation has received a lot of attention from researchers. Traditionally, sensors are mounted on robots to detect the surroundings. However, they are sometimes inaccurate due to the pertinent problem of dealing with uncertainty in the environment. Fuzzy logic has long been regarded as a useful method for dealing with ambiguity that arises from imprecise knowledge. For wheeled mobile robots, several researchers have proposed three input proximity sensors and fuzzy logic controllers with 27 rules. Although this approach is interesting, it fails to account for uncertainty, resulting in difficulties avoiding obstacles. This research aimed to improve a Mamdani fuzzy controller by increasing the number of sensors, reducing the number of fuzzy rules, and revising membership functions. The developed type-1 fuzzy controller (M) was then compared to its corresponding type-2 (K). Simulation research method was adopted using commercially available V-REP and MATLAB software. A purposive sampling technique was chosen, and all simulations were run fourteen times. The results presented a new Mamdani fuzzy logic model with nine inputs, two outputs, and eighteen rules. ANOVA test revealed a significant effect of membership functions at p<.05 level for the three conditions [F(2, 39) = 9.17, p = 0.001]. Model comparison was done using an independent samples t-test. Model K had a higher score (M = 219.79, SD = 4.509) than model M (M = 223.79, SD = 3.886), indicating a disparity in the models t (26) = 2-514, p = 0.018) with a Cohen's d effect size of 0.924 meaning a large effect. Triangular membership functions constitute an immediate solution to the optimization problems in fuzzy logic modeling. Type-2 overcomes the limitations of type-1 fuzzy controllers presenting a way forward to fuzzy controllers in highly uncertain environments and real-world applications. It is envisaged to see a widespread use of type-2 fuzzy logic controllers in the next decade. The prospect of the thesis catalyzes additional research into a hardware implementation. This will aid in the development of robots for use in hazardous and crowded environments.