100 likes | 196 Vues
This project aims to build multi-classifiers to classify different swarm behaviors like Disperse, Orbit, Cluster, and Bubble Sort based on raw data and pattern recognition algorithms. Initial results show promising classification accuracy with KNN, Neural Nets, and Bayes Nets. Further fine-tuning and experimentation are needed for improvement. Future directions involve incorporating temporal information for learning hidden emergent sub-behaviors.
E N D
Aisha Walcott-MAS622J-Dec. 11, 2006 Robot Recognition of Complex Swarm Behaviors
Introduction A Swarm is a large collection of autonomous mobile robots No centralized control Group behaviors are produced from local interactions of many individual robots Goal is to develop a suite of primitive global behaviors that combine to form more complex group programs Dispersion Orbit Courtesy James McLurkin
source * source Project Goal Build multi-classifiers to classify Complex Swarm Behaviors Disperse Orbit Cluster Bubble Sort Example Features hi source low source
Approach • Collect raw behavior data sets • Determine Features (8D) • Pattern Recognition Algorithms • KNN • Neural Nets • Bayes Nets • Analyze results of each algorithm
KNN • Tested a range of values for nearest neighbors random tie break Overall Correct Classification Average Class Classification Cluster= 100% Disperse = 12.5% Clump = 50% Orbit = 44% Bubble Sort = 82%
Single Hidden Layer Layer 1: nodes [50,70] Max percent = 65% Logsig Two Hidden Layer Layer 1: nodes [50,70] Layer 2: nodes [25,25] Max percent = 65% Neural Nets
Bayes Nets • Mapping to discrete domain by applying k-means clustering to each feature Preliminary Results Classification of Cluster Possible bug in code Modify the discrete mapping Cluster, Disperse,Clump,Orbit, Bubble Sort 8 1 2
Discussion • KNN and Neural Net performed well • Determining the mapping from real numbers to a discrete domain may affect Bayes Nets classifiers • Overall high classification of clustering behavior • -Features tuned to behavior • -Not enough variety of samples • Need more samples of varying behavior
Next Steps • Feature selection-which group of features work best for each classifier • Additional experiments to determine why certain classifications are much better • Future • Use the temporal information to learn hidden emergent sub-behaviors