Exploiting Evolutionary Algorithms for Improved Automatic Target Acquistion

The focus of this SBIR Phase I program is to develop and test a feature-based automatic target acquisition (ATA) system exploiting evolutionary algorithm (EA) technology. The ATA system will accept phenomenologically-based target and scene featuresextracted from high resolution LADAR imagery and output a target classification decision with a high probability of correct classification. The use of LADAR data is particularly attractive since it provides 3D imagery while simultaneously achieving theangular resolution of I2R and range resolution of SAR at a lower cost than either of those two sensors. The EA approach represents an innovative structure for both improving ATA performance and streamlining computations associated with model-basedtechniques. The approach utilizes proven evolutionary algorithm software that has been developed and successfully implemented on other ATA systems.The results of this research will broaden the technology base of advanced ATA algorithms available to the Air Force as it continues to pursue low-cost, autonomous LADAR seekers for conventional weapons guidance. The performance of the EA classifier willbe benchmarked against existing model-based approaches with the metrics being run time, memory requirements, and percent correct/incorrect classification.The automatic vehicle recognition and identification capabilities emanating from this SBIR will have a number of dual-use applications. For instance, the results would be pertinent to aircraft tracking and control at commercial airports, intrudertracking, and intelligent highway applications. Automated systems involving manufacturing inspection (including robotic bin picking) would also benefit from innovative recognition and identification algorithms.