ECSEL 2017 RIA aFarCloud Qamcom
| Reference number | |
| Coordinator | Qamcom Research and Technology AB |
| Funding from Vinnova | SEK 1 213 496 |
| Project duration | September 2018 - November 2021 |
| Status | Completed |
| Venture | ECSEL |
Important results from the project
In this project, we aimed to develop a ML platform for automatic detection of cattle in drone images. Combined with the possibility to geolocalise the detected animals, this supports farmers in their daily inspections. In parallel, we developed a methodology for hazard analysis and risk assessment for an autonomous system of drones. This work mainly focused on the safety analysis of a UAV system engaging in autonomous drone-cloud based missions such as crop sampling, herd health monitoring, etc. Both goals have been achieved with satisfactory results.
Expected long term effects
Qamcom´s machine learning platform represents the computational backend in an integrated system comprising a mission management tool and a second image processing platform geolocalising the detected animals (developed by project partners). This end-to-end integration was successfully demonstrated at the end of the project. As for the safety analysis of a UAV system, we combined two safety standards, ISO 12100 and ISO 13849, the former used for risk assessment and the latter to determine the performance level of the safety functions specified for the UAV system.
Approach and implementation
The work followed the subdivision in work packages imparted by the project. In particular, we were active participants of work packages 2 (System Requirements, Architecture & Specification, and Implementation), 4 (Environment characterization platform), 6 (Autonomous System Development and Legacy System Integration), 7 (Demonstrators definition, integration, verification and validation) and 8 (Innovation Management, Dissemination, Exploitation and Standardization).