Comparison of Fuzzy Optimization and Genetic Fuzzy Methods in Solving a Modified Traveling Salesman Problem

Mitchell, S.; Ernest, Nicholas; Cohen, Kelly

doi:10.2514/6.2013-4664

Cited by 3 publications

(2 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Collaborative Learning using Fuzzy Inferencing (CLIFF) [2] uses type-2 logic to create an intelligent and dynamic robotic coach. Precision Route Optimization using Fuzzy Inferencing (PROFIT) [3] [4] examines the use of fuzzy logic to create an optimal result in a traveling salesman problem.…”

Section: Introductionmentioning

confidence: 99%

Fuzzy logic decision making for autonomous robotic applications

Mitchell

Cohen

2014

2014 IEEE 6th International Conference on Awareness Science and Technology (iCAST)

Self Cite

View full text Add to dashboard Cite

There is growing in interest in the effectiveness of emulating human decision making and learning in modern aerospace applications. The following is an examination of several applications in which type 1 and 2 fuzzy logic has been utilized in artificial intelligence and machine learning problems to demonstrate their capabilities. In Fuzzy Logic Inferencing for PONG (FLIP), the effectiveness of type 1 logic is examined as an optimal controller for players in the game of PONG. Robotic collaboration is also developed as the PONG game was expanded into a multiplayer option. Collaborative Learning using Fuzzy Inferencing (CLIFF) is an extension of this PONG game, however type-2 logic is used to create a robotic coach that optimizes its players to beat its opponent in a development of layered fuzzy learning. Precision Route Optimization using Fuzzy Intelligence (PROFIT) examines the use of fuzzy logic as an optimizer in an algorithmic solution to a modified Travelling Salesman Problem (TSP). The TSP is modified in a way to better mimic a real-lifescenario where footprints must be visited instead of simply points, which gives an interesting complexity to the problem. Considering the successes associated with these research endeavors, it can be concluded that type 1 and 2 fuzzy logic are both interesting tools that can further the abilities of intelligent systems and machine learning algorithms.

show abstract

Section: Introductionmentioning

confidence: 99%

Fuzzy logic decision making for autonomous robotic applications

Mitchell

Cohen

2014

2014 IEEE 6th International Conference on Awareness Science and Technology (iCAST)

Self Cite

View full text Add to dashboard Cite

show abstract

“…The genetic-based PSO procedure is then applied to solve the TSP with better efficiency in the second phase. Mitchell et al [15] used fuzzy optimization of a path, produced using GA, to solve the TSP. The project was made more realistic by using Dubins paths and finally comparing this to genetic fuzzy algorithm to determine differences in accuracy and precision between the two solutions.…”

Section: Introductionmentioning

confidence: 99%

Comparison of Approximate Approaches to Solving the Travelling Salesman Problem and its Application to UAV Swarming

Sathyan¹,

Cohen²,

Boone³

2015

IJUSEng

Self Cite

View full text Add to dashboard Cite

Sathyan A, Boone N and Cohen K. (2015). Comparison of approximate approaches to solving the Travelling Salesman Problem and its application to UAV swarming. International Journal of Unmanned Systems Engineering. 3(1): 1-16. The Travelling Salesman Problem (TSP) is a widely researched Non-deterministic Polynomial-time hard optimization problem with a range of important applications in a wide spectrum of disciplines including aerospace engineering. In this paper, a comparison of different approaches to solve the TSP and also its application towards swarming of UAVs is considered. The objective of the TSP is to determine the optimal route associated with the shortest tour connecting all targets just once. Genetic Algorithms (GA) are one of the most widely applied techniques for solving this class of optimization problems. Two other techniques, 2-opt and Particle Swarm Optimization, are used and the results are compared with those obtained using GA. The comparison is made for different numbers of targets, using salient figures of merit such as computational time required and the cost function which is the minimum solution (distance) obtained. Results show that the 2-opt approach with the closest neighbour as initial starting point for the search yields superior performance. In the Multiple Travelling Salesman Problem, we propose a cluster-first approach which allocates each specific UAV to a subset of targets. The 200 targets are divided into four clusters corresponding to the four UAVs and then TSP algorithms like 2-opt and GA are employed to solve each cluster. This approach drastically reduces the computational time and also gives much better results than the conventional technique of directly applying GA over the 200 targets.

show abstract