Description
The objective of this activity is to design and perform an in-orbit experiment investigating autonomous deployment and early operations for telecom constellation satellites. This includes the development of command and data-handling hardware as well as AI/ML based operational algorithms developed on-ground.Targeted Improvements: Increase of onboard autonomy by 50% (i.e. include the key number of autonomous functions), including attitude acquisition, deployable actuation, establishment of communication links, early manoeuvres.Description: Future satellite constellations will need an advanced level of autonomy, with more than 40,000 satellites are forecast to be launched in the next 10 years. Traditional operations of these satellites will not be viable, and it is expected that the space and ground infrastructure will need advanced autonomy capability. With this increasing number of satellites in low Earth orbit, there is a need to reliably dispose of the satellites at end-of-life to avoid on orbit collisions, which requires health monitoring capability to support end of life management. To enable future satcom constellations, the issues of operational overhead and compliance to clean space requirements need to be addressed. Recent advances in AI/ML algorithms and decentralised intelligence techniques have the potential to analyse and interpret all the available platform telemetry, allowing it to make decisions and adjust itsoperations in real-time. Autonomy reduces the need for constant ground control and enables satellites to respond to changing conditions or unexpected events. Artificial (AI) / Machine Learning (ML) algorithms are able to be employed to detect anomalies and diagnose faults in satellite systems. By analysing telemetry data and comparing it with known patterns, AI models can identify potential issues in real-time, enabling predictive failure tools to update the platform lifetime and to initiate end of life procedures beforecritical platform failures make this impossible.This activity will design, develop and utilise a flying testbed for testing AI/ML enabled autonomous operations. The in-orbit test bed will select and embark a leading-edge AI hardware accelerator to process and reactto received telemetry. The in-orbit test bed will be deployed in an operational environment for at least 12 month and be capable ofperforming critical operations during the deployment and early operations phases.Specifically the in-orbit test bed will:- EmbarkanAI enabled onboard controller that has access all available telemetry, sensor data and control the actuators onboard the satellite.- Host operational algorithms for deployment, early operations and end-of-life procedures.- Allow processing with and withoutAI/MLtechniques, allowing AI performance improvements to be quantified.- Allow the in-orbit upload of additional/modified operationsalgorithms. The baseline experiments will:- Simulate the automated deployment of antennas, solar arrays, initiation of beaconing and establishment of first ground contact.- Adapt to fault injection (sensor noise/failure, actuator degradation/failure)- Benchmark traditional vs AI based Autonomous Position Correction System (APCS) and Guidance, Navigation and Control (GNC) building blocks (which will be crucial e.g. to implement autonomous collision avoidance)Footnote: On Delegation Request activities will only be initiated on the explicit request of at least one National Delegation.