Simulation-Based Results of the Trusted Distributed Autonomy Demonstration Experiment

This distributed autonomy solution was developed and tested to demonstrate fully autonomous, distributed, trusted, and adaptive onboard autonomy capabilities that enable resilient management of resources in proliferated space constellations to meet emerging needs. The work we describe is an integrated distributed autonomy software solution developed jointly with the United States Air Force Research Laboratory. The autonomy software was implemented for demonstration in a physical hardware environment. We present results from demonstrations of two key complimentary autonomy software components that work in tandem: a distributed information fusion and target tracking technique called the distributed Particle Gaussian Mixture Filter (dPGMF) and a resource management capability, based on Artificial Intelligence and Machine Learning (AI/ML), called distributed Shielded Deep Reinforcement Learning (dSDRL).