Trusted Autonomy

Overview

  • Multi-satellite swarms and constellations are quickly becoming the norm in civil, military and commercial space. These systems need to perform complex and coordinated maneuvers such as cooperative data collection, and rendezvous and proximity operations.  Furthermore, they need to stay safe by reacting quickly to faults and external threats.

  • Distributed, mission-level autonomy is a key enabler for complex and large-scale multi-satellite missions, but the mission operators need to be able to trust the autonomy to do the right thing.

Emergent Application

  • Emergent is developing a suite of SpaceApps, collectively called Commander, that perform distributed, mission-level autonomous commanding of multi-satellite swarms and constellations 

  • Commander focuses on technologies and algorithms that facilitate transparency, ease of understanding, and supervisory control over mission execution

  • Commander interfaces with Emergent’s other SpaceApp suites for cluster management (Autopilot), fault detection, isolation, and recovery (Guardian), and space cloud computing (Cirrus) to enable advanced autonomous multi-satellite mission capabilities

Trusted Autonomy through…

Mission monitoring and supervisory control

Ascent Simulation Visualization

  • Ground operators need visibility into, understanding of, and supervisory control over the autonomous system at all levels (mission, agents, and tasks)

  • Emergent is developing graphical user interfaces (GUIs) for Commander that enable mission operators to view telemetry and mission status, edit and upload new plans, and issue supervisory control commands to intervene when necessary

Trusted Autonomy through…

Integration with Emergent’s simulation and testing tools

Summit CI Test Interface

  • Integration with the ASCENT modeling, simulation, and analysis framework allows an operator to inspect the outcome of a new plan before tasking real assets

  • Integration with the SUMMIT CI/CD test environment enables continuous integration testing of a library of plans under different conditions

  • Frequent testing before and during missions enables operator confidence in the autonomous system

Trusted Autonomy through…

Clearly defined interfaces and roles

Commander Space Architecture

  • Commander facilitates a clear division of responsibility between the operator, mission-level, agent-level, and task-level execution services

  • Commander software can be run on-board or from the ground, with the operator in-the-loop or on-the-loop, thereby enabling an evolutionary approach to trusted autonomy 

  • The operator interacts with Commander by issuing commands, monitoring telemetry, and uploading new plans