ENORMITY: pErformance beNchmark Of Ramon space nustreaM wITh klepsYdra ai
ENORMITY: pErformance beNchmark Of Ramon space nustreaM wITh klepsYdra ai
批准号:
10089138
负责人:
金额:
$8.78万
依托单位:
依托单位国家:
英国
项目类别:
Collaborative R&D
财政年份:
2024
资助国家:
英国
项目状态:
未结题
起止时间:
2024 至 --
中文摘要
点击翻译按钮获取中文摘要
英文摘要
The industrial research project ENORMITY shall investigate enhanced AI-based data processing on-board satellites and spacecraft. Such data processing requires on one hand radiation hardened (rad-hard) Space computer and storage systems and on the other hand AI software to efficiently process the data on this computer hardware.The combination of the high-performance rad-hard space computer from Ramon.Space (UK) with the high-performance AI software from Klepsydra (CH) (in collaboration with ETH, CH) will result in new capabilities of satellite systems, such as on-board feature detection and cognitive instrument for Earth Observation satellites, AI based navigation and landing capabilities for planetary missions, and increased autonomy for In-Orbit Service missions. ENORMITY will enable Space systems to benefit from capabilities AI can provide to autonomous and remote-controlled satellites and spacecraft. Today, satellites cannot fully benefit from AI, as either AI accelerator Integrated Circuits (ICs) are unable to withstand the harsh radiation conditions in space, causing processing errors and re-boots of on-board computers and terminally failing after short lifetime, or the available rad-hard processing ASICs are difficult to program due to their very specific architectures.ENORMITY will for the first time combine an easy to program and easy-to-use software (SW) framework with powerful rad-hard Space-resilient processors. ENORMITY addresses the growing market of Earth Observation services as well as the nascent market of commercial lunar missions. In addition, the ENORMITY Hardware / Software (HW/SW) combination will be very useful for future scientific missions for ESA, NASA, JAXA and other Space exploration agencies. ENORMITY also opens the path for new capabilities such as enabling Vision Transformer AI models using the same HW/SW architecture.
期刊论文(0)
专著(0)
科研奖励(0)
会议论文
海外基金