Matthew Nelson
Charles River Analytics collaborated with scientific consulting firm Metron and three universities to create approaches that will facilitate adaptation of unmanned underwater vehicle software to changes in the platform's internal and external environments.
The Probabilistic Representation of Intent Commitments to Ensure Software Survival effort was part of the Defense Advanced Research Project Agency's Building Resource Adaptive Software Systems initiative and sought to produce adaptable software systems through machine learning and probabilistic programming techniques, Charles River Analytics said Wednesday.
Academic partners for PRINCESS were University of Southern California, Harvard University and University of Birmingham.
The team used the REMUS 600Â autonomous underwater vessel to demonstrate the UUV's acclimation to new architectures, system failures as well as downgraded and updated sensors.
