Jun 12, 2014  
Fuzzy logic brings robots closer to humanlike reasoning 

(Nanowerk News) University of Cincinnati student Sophia Mitchell’s latest paper, Precision Route Optimization using Fuzzy Intelligence, is getting international attention with her invitation to the International Conference on Awareness Science & Technology in Paris, France. While Sophia’s international recognition is particularly significant as an undergraduate, this is not her first time.  
“As a freshmen in high school I was the youngest student from Kentucky to win the state science fair and become eligible to compete in the Intel International Science and Engineering Fair (ISEF) … The next year, I won overall, making me eligible to compete in ISEF (again) where I won fourth place in my category for a continuation of the same research (heliophysics).”  
Sophia says even though her latest research is in an entirely different field from her work in high school, she still feels the same sense of gratitude that come from creating something others find exciting and meaningful.  
Mitchell’s project, PROFIT (Precision Route Optimization using Fuzzy Inference) is one of three projects which she began while working with Professor Kelly Cohen and Professor Manish Kumar, who is now at the University of Toledo. PROFIT falls under the Ohio Space Grant Consortium (OSGC), which aims to advance the nation’s capability in STEM (Science, Technology, Engineering, and Mathematics) through a comprehensive scholarship and fellowship program.  
Sophia’s latest research comes from her longtime interest in the interworkings of the human mind. She says, “Programming robots that can reason, collaborate, create and learn like humans creates an interesting opportunity to look into how our brains work.”  
Her research on fuzzy logic robots with humanlike reasoning is more than just fun to talk about. According to Mitchell, real world applications for robots programed with fuzzy logic include improving space robotics, disaster relief, and military efforts.  
In Sophia’s paper she explains, “Normally in programming, one would use binary logic, where things are either zero or one, on or off, black or white. In real world decision making, using a binary system for decision making wouldn’t work very well, as nothing in the universe is completely one thing OR another, but on a continuum.” Fuzzy logic works with the continuum and allows robots to make decisions that are not black or white. This type of logic accounts for the gray areas in decision making, creating more humanlike reasoning in robots.  
Mitchell’s paper focuses on the “Traveling Salesman Problem” (TSP). She explains this problem answers the question, “given a set of points in a sample space, what is the optimal route that touches each point in the sample space before returning to the starting position. Here, an optimal route is generally defined as the fastest or shortest route between two targets.”  
Each “target” is a physical point that a robot must reach in order to complete the TSP. Traditionally, TSP targets are all points, however, in Mitchell’s research she turns these points into areas, which is slightly more general. This is more realistic, as a traveling robot generally only has to reach a certain area in order to complete its task, whether it be collecting data from a radio footprint or taking pictures. Traditional methods would be overwhelmed in processing areas.  
Currently there are two methods used to find solutions for TSP. The first method uses a numerical algorithm, or complex formula to compute exact solutions to the problem. The flaw with this method is that it only functions well with small problems, defined as a small number of targets, making it ineffective for real life problems, which normally include a large number of targets. The second method uses a genetic algorithm, which Mitchell compares to “the process of how humans think about breeding animals.” She explains, “Before the algorithm runs, the ‘best solution’ from the starting values are chosen to create the next solution set... Mutations can occur, where the algorithm randomly places values in the solution space.” This approach is experience based and effective when solving small problems but loses its optimization as more targets are added. It also requires a lot of computational power as the problems get larger and more complex, which means applying a genetic algorithm to a TSP where targets are areas would take some time to compute. Mitchell improves this second method genetic algorithm with the creation of a fuzzy algorithm. 

The fuzzy logic is an added layer on top of the results of a genetic algorithm. The genetic algorithm finds the best path between the center points of the target areas, then Sophia’s layer of fuzzy logic considers the entire footprint around these targets and creates a better path based on these areas.  
This method is better than what is possible with the previous two. Not only does it not require experience to create an optimal path, it also is an extremely quick method.  
Mitchell's creation is both robust and can be used in timecritical situations.  
Essentially, the system works by a genetic algorithm solving a simplified version of the TSP, then Mitchell’s fuzzy logic layer adds in complexity and optimizes the solution (i.e. Determines the best possible route for the robot).  
Her research is successful decreasing the path distance between targets by at least 5 percent in most cases.  
The success of Sophia Mitchell’s research is just one of many. Her passion for robotics and space has pushed her to do great things since age 6, when she first began her dreams of being an astronaut. After teaching herself statistics and complex physics in the 6th grade, Mitchell began conducting independent research and eventually went on to win numerous science fairs in high school. She was also given the opportunity to preform research at the university level, conducted with a postdoc at the University of Louisville on astrophysics. That work has since been published in the Monthly Notices of the Royal Astronomical Society.  
In 2012, as a prejunior in aerospace engineering and an ACCEND student, Mitchell received the “Best Presentation” award at the 7th annual Dayton Engineering Sciences Symposium for FLIP, Fuzzy Logic Inferencing in PONG. The Ohio Space Grant Consortium (OSGC) has also recognized Mitchell for her other research project, Collaborative Learning using Fuzzy Inferencing (CLIFF).  
Mitchell would like to give a special thanks to Professor Kelly Cohen for being, “quite simply the world’s greatest adviser.” She explains, “He took the chance of taking me onto his research team as a freshman, and has been a great source of guidance and knowledge since then.”  
Mitchell hopes to represent the University of Cincinnati in France at the International Conference, where she is sure to stand out and make her city proud. 
Source: University of Cincinnati  
Subscribe to a free copy of one of our daily Nanowerk Newsletter Email Digests with a compilation of all of the day's news. 
These articles might interest you as well: