Artist’s rendition of a satellite – mechanik/123RF Stock Photo
This week three separate reports caught our attention about how Artificial Intelligence (AI) is impacting the space and Earth Observation (EO) communities. This covers both upstream with AI controlling satellites, and downstream activities with data applications.
China Using AI To Control Satellite
For the EO community, there was an interesting article from China over the weekend in the South China Morning Post. According to a paper published in the journal Geomatics and Information Science of Wuhan University, Chinese scientists allowed an AI to take full control of an EO satellite for 24 hours.
The satellite, Qimingxing 1, was originally launched on the 27th February 2022 by the Long March-8 Y2 rocket on behalf of Wuhan University. It is small hyperspectral EO satellite with 32 spectral bands, and has a spatial resolution of 21 metres in the visible spectrum mode and 100 metres in infrared mode, with both modes having a swath width of around 100 kilometres.
The experiment undertaken was to leave the AI to manage the satellite without any human intervention for 24 hours. According to the report, the AI selected several places and tasked the satellite to capture images, including:
- Patna city in India
- Osaka region of Japan.
It is not clear why the AI picked these areas, although it is worth noting that the Patna is the home of the Bihar Regiment of the Indian Army, and Osaka is a busy port and occasionally hosts US Naval vessels in the area. The researchers were surprised how the AI made decisions and managed the satellite, calling the work a ‘ground breaking experiment’.
There are clearly lots of questions arising from this experiment in terms of how the AI was trained and how it was making its decisions. However, it is interesting, for EO data providers that such a system can be used and whether it could potentially make more efficient and effective use of the satellite than human interaction.
Ireland Using AIÂ & Earth Observation To Monitor Peatlands
A new project has been launched in Ireland to determine how AI can help ‘monitor and protect’ the country’s peatlands, and increase biodiversity through peatland restoration. The AI2Peat Project, is a collaboration between the National Parks and Wildlife Service, CeADAR (Ireland’s National Centre for Applied AI) and iCRAG (Science Foundation Ireland Research Centre in Applied Geosciences).
Peatlands cover around 17% of Ireland and this project aims to use remote sensing data from both satellites and drones, combined with citizen science on ground data collection, to develop a map of peatland areas under ‘threat from erosion, exploitation and climate change’ together with areas of high conservation value. The project also aims to influence environmental protection, biodiversity, and climate change policy in the country.
The project is one of 26 taking part in a €65m National Challenge Fund competition to investigate how emerging digital technologies can be used to help Ireland achieve its goal of becoming carbon neutral by 2050. The current phase is a six-month concept phase with €50,000 funding, and successful projects will go onto a seed phase receiving €150,000, and ultimately the selected project will receive €500,000. It will be fascinating to follow the progress of this project in the coming months.
United States Using AI For Earth Observation Land Classification
Finally, it was reported last week that the American company Impact Observatory has raised $5.9 million in a seed round of financing. Amongst other aspects, the money will be used to accelerate the launch of its AI remote-sensing monitoring products. Impact Observatory combines Sentinel-2 and Landsat imagery with its own machine learning to deliver real-time automated land use and land cover classifications; although the company hopes to now add commercial datasets into the solution. The resulting dataset is to be used for applications such as agriculture and food security, monitoring carbon emissions, climate change, and protecting biodiversity and ecosystems.
Conclusion
There is no doubt that AI, in all forms, is going to be part of the biggest change in both business and jobs in the coming years. The organisations that recognise this now and harness the potential will be the ones that thrive, and all EO companies need to take note and consider their next steps for using AI. However, the human should not be forgotten as these systems rely on training and guidance as seen by the recent releases of ChatGPT and similar services.
