In space exploration, time moves quickly—and so does Intuition-1. Launched November 11, 2023, our 6U satellite has now completed its first full year, orbiting Earth 5490 times and showcasing how advanced artificial intelligence and onboard data processing can transform space-based Earth observation. As a technology demonstrator, Intuition-1 has consistently proven its capacity to redefine the field. Let’s revisit the technical milestones that have marked this mission over the past year.
In-Orbit Data Processing: The Core of Intuition-1’s Capabilities
What truly defines Intuition-1’s success is its on board Leopard Data Processing Unit (DPU). This unit isn’t just another component; it’s the satellite’s central processing powerhouse, enabling Intuition-1 to perform data analysis while still in orbit. This advanced processor can execute up to three trillion operations per second and uses artificial intelligence (AI) and machine learning algorithms to analyze hyperspectral data autonomously.
The Leopard DPU supports tasks such as cloud detection, image segmentation, and data classification, allowing only the most relevant information to be sent back to Earth. This approach optimizes the satellite’s bandwidth usage and reduces the volume of raw data needed, enabling timely insights for applications such as environmental monitoring and disaster response.
Over the past year, the Leopard DPU has processed terabytes of data from the hyperspectral optical instrument. Using FPGA-accelerated deep-learning inference, the DPU has successfully performed in-orbit tasks like cloud masking, supported by both classical and AI-driven image-processing algorithms. These include feature points-based data co-registration, radiometric correction, hypercube rendering, and segmentation models like KP Labs’ Deep Earth convolutional network and U-Net-based algorithms. With these capabilities, the Leopard DPU is paving the way for space-based data processing, delivering a new level of autonomy and efficiency.
Moreover, the DPU’s parallel processing capabilities have been tested extensively in orbit, demonstrating its ability to handle heavy-duty computations flexibly. The system’s redundancy in processing nodes ensures that essential tasks remain uninterrupted, further supported by error-correction mechanisms that have consistently prevented single-event functional interrupts (SEFIs) over the past year.
Hyperspectral Imaging: A New Benchmark in Earth Observation
While the Leopard DPU is the operational heart of Intuition-1, the satellite’s hyperspectral optical instrument supplies the essential data it processes. This sophisticated sensor captures data across 192 narrow spectral bands, from visible to near-infrared wavelengths, offering a much richer spectral view of Earth’s surface than conventional satellites using only three broad RGB bands.
Since its deployment, Intuition-1 has delivered an array of hyperspectral images from regions including Chad, Turkey, Alaska, Saudi Arabia, Algeria, Poland, Brazil, and Spain. Notably, an image over Utah Lake in the USA highlighted vegetation density, with the Leopard DPU performing real-time image segmentation and classification to provide ecological insights using both visible and hyperspectral bands.
Enhanced Data Accuracy through Cloud Detection
One of the potential applications of the Leopard DPU involves cloud detection. Using a spatio-spectral AI model, the DPU identifies cloud-covered areas, enabling the selection of clear-sky images for transmission back to Earth. This capability contributes to improved data accuracy and optimized bandwidth usage, enhancing Intuition-1’s overall mission efficiency.
Thermal Management and System Resilience
Space presents numerous environmental challenges, and effective thermal management is critical to mission success. Over the past year, Intuition-1’s systems, particularly the Leopard DPU, have demonstrated exceptional resilience. The DPU’s stable thermal performance is maintained by an advanced cooling system designed by KP Labs, which utilizes conduction cooling to transfer heat to external radiators, dispersing it into the cold vacuum of space. This setup allows the DPU to maintain optimal temperatures, even during extended high-performance data processing sessions.
In addition, the Leopard DPU’s error-correction mechanisms have successfully mitigated SEFIs, ensuring continuous mission operations without interruption. KP Labs engineers rigorously tested and validated these systems during the satellite’s development, preparing Intuition-1 for the demanding conditions of space.
What’s next?
Intuition-1 has proven to be much more than a technology demonstrator; it is a robust example of autonomous, space-based data processing. With the Leopard DPU at its core, Intuition-1 continues to reveal the vast potential of AI-driven onboard data processing. The satellite’s ability to capture and process high-resolution hyperspectral images in orbit has yielded valuable insights into diverse environments, from industrial areas in Poland to the biodiverse ecosystems of the Amazon.
We believe that the experience we gained during this mission will further underscore the critical role of the Leopard DPU in harnessing the full potential of AI and hyperspectral imaging within the Earth Observation sector.