Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2022, Vol. 19, No. 1, pp. 107-122
Tasks of normative and technical regulation of intelligent Earth remote sensing systems
1 National Research University Higher School of Economics, Moscow, Russia
Accepted: 04.03.2022
DOI: 10.21046/2070-7401-2022-19-1-107-122
The article deals with the main Earth remote sensing (ERS) tasks that are solved using artificial intelligence (AI) technologies. A task classification is singled out which is backed by the universal classification of applied intelligent tasks based on functional similarity of the artificial and natural intelligence. The article shows the types of data for which processing the use of the AI methods is most reasonable. The main regulatory and technical barriers have been defined preventing from efficient design and usage of the AI systems in ERS as well as standardization objectives intended to break these barriers. When setting up a list of crucial standardization objectives, the global and national experience in the AI standards development for ERS has been taken into account. To make the suggested objectives list complete, an approach is offered based on the holistic analysis of the ERS intelligent system life cycle processes. Therefore, the following are suggested as the main groups of tasks for regulatory and technical solution of AI usage problems for ERS: validation of requirements and harmonization of procedures for measuring essential functional characteristics of the ERS intelligent systems (ERSIS); assessment of functionalities of a skilled human operator solving a certain ERS application task manually; formalization of the specified ERSIS operating conditions; additional ERSIS learning control during operation, replication of designed software-based algorithmic decisions on related ERS tasks; provision of data confidentiality when designing and applying ERSIS; harmonization, assurance of quality and increase of accessibility of data sets required for ERSIS development and conformance evaluation.
Keywords: artificial intelligence, applied problems of artificial intelligence, intelligent problems of remote sensing of the Earth, system life cycle, evaluation of the functional characteristics of intelligent systems, intellimetry, quality of artificial intelligence systems
Full textReferences:
- Garbuk S. V., Bakeev R. N., Competitive assessment of the quality of intelligent data processing technologies, Problemy upravleniya, 2017, Vol. 6, pp. 50–62 (in Russian).
- Garbuk S. V., Gubinskii A. M., Iskusstvennyi intellekt v vedushchikh stranakh mira: strategii razvitiya i primenenie v sfere oborony i bezopasnosti (Artificial intelligence in the leading countries in the world: development strategies and application in the field of defense and security), Moscow: Izd. “Znanie”, 2020, 599 p. (in Russian).
- Chen L., Zhu G., Li Q., Li H., Adversarial Example in Remote Sensing Image Recognition, arXiv preprint, arXiv:1910.13222v2, 2020, 12 p., available at: https://arxiv.org/pdf/1910.13222.pdf.
- Czaja W., Fendley N., Pekala M., Ratto C., Wang I., Adversarial Examples in Remote Sensing, Proc. 6th ACM SIGSPATIAL Intern. Conf., 6–9 Nov. 2018, Seattle, WA, USA, 2018, arXiv: 1805.10997v1 [cs.CV].
- Garbuk S. V., Intellimetry as a way to ensure AI trustworthiness, Proc. 2018 Intern. Conf. Artificial Intelligence Applications and Innovations (IC-AIAI), 6–10 Oct. 2018, Limassol, Cyprus, 2018, pp. 27–30, DOI: 10.1109/IC-AIAI.2018.00012.
- Novikov G., Trekin A., Potapov G., Ignatiev V., Burnaev E., Satellite Imagery Analysis for Operational Damage Assessment in Emergency Situations, In: Business Information Systems (BIS 2018): Lecture Notes in Business Information Processing, Abramowicz W., Paschke A. (eds), Cham: Springer, 2018, Vol. 320, pp. 347–358.