When accessible, the project description wrote that ”COMPASS2020 aims to demonstrate the combined use and seamless coordination of manned and unmanned assets to achieve greater coverage, better quality of information and shorter response times in maritime surveillance operations. By combining innovative technologies and integrating them within the current operational procedures, COMPASS2020 solution ensures long range and persistent surveillance, increasing the situational awareness of Coast Guards and maritime authorities, thus increasing the cost-effectiveness, availability and reliability of the operations.”
Interestingly, this is framed as a solution to a problem that limits the current uses of surveillance: “Although there has been an expressive investment done in the domain of surveillance technologies and tools, the intake by the competent authorities has been slow, due to lack of uniformity in the integration of such systems with existing surveillance infrastructures.”
COMPASS2020 is also defined as a “system of systems” on its Cordis page, providing at the same time 1) a human-machine interface for situational awareness, 2) a data fusion platform to enhance decision-making, 3) a bundle of new services, including “anomaly detection, classification, risk & predictive analysis and optimisation of operations with unmanned platforms, e.g. automatic tasking.”
Video presentations — including animated infographics — are available on the project’s YouTube channel, where we can read its motto: “expanding surveillance reach.” Otherwise, a dramatic video intro asks: “Who will protect you?”
The project’s main technological outputs include:
1) UAVs with “improved surveillance capabilities” (described here, even though it is not clear what these “improved capabilities” actually are);
2) a central, interoperable, “multi-domain” “Mission System”;
3) data fusion;
4) threat analysis (cfr. the project’s ppt presentation).
The project’s brochure further states that “Compass Mission System uses artificial intelligence techniques to complement the system overall capabilities. The system can automatically determine suspect and irregular behavior and focus the attention of the operator to this situation. It provides advise on the best way of intervention. The COMPASS 2020 Artificial Intelligence reduces the operator workload, enabling the persistent surveillance of extended coastal areas.”
COMPASS2020 is explicitly portrayed as a follow-up to a previous EU-funded project. “It is to be mentioned that COMPASS2020 can be treated as a kind of follow up of the SWARMs (Smart and Networking UnderWAter Robots in Cooperation Meshes) project, whose main objective was to develop an integrated platform for autonomous maritime and underwater operations including AUV, aerial autonomous vehicles (AAVs), ROVs and unmanned surface vehicles (USV) performing dangerous missions (…) in multimodal modes.”
The projects also enjoyed a collaboration with the EFFECTOR and ANDROMEDA projects, and was deemed “of interest” by Frontex in 2020.
Given the jarring absence of public deliverables — only two are available on Cordis at the time of writing, i.e. almost three years after project completion — there are no detailed information on tests/demos for project outputs.
We however know that “The demonstration as the conclusive project event will be conducted in Portuguese waters by Autoridade Marítima Nacional agency in Portugal conducting surveillance activities in domains of policing and security duties. The scenario event will consist of pre-organised situations of narcotics smuggling and the search and rescue (SAR) of irregular immigrants”, writes a paper produced as a deliverable for the project (and marked “NATO UNCLASSIFIED RELEASABLE”).
It is interesting to notice how the project fundamentally bundles narcotics smuggling and irregular migration together: “COMPASS2020 project has been conceived to reduce the number of illegal border crossings through improving coordinated actions supported by manned and unmanned (underwater, sea surface and air) vehicles (UxV). Besides struggling with illegal migrants (“illegal” rather than “irregular” being an interesting word choice in itself, ed.), EU has a problem related to narcotics trafficking. (…) Consequently, COMPASS2020 was also conceived to combat this issue.”
From deliverable D6.3, which is publicly accessible through Cordis, we also learn that the Consortium deemed the results of such demonstration satisfying: “Overall COMPASS2020 demonstration was a success in all means: demonstration of technical capabilities of all systems involved, personnel coordination, logistics management, team work between different partners and involvement of end-users in validation of the system performance”, writes the deliverable.
But even though “All components have been tested in operational environment”, there was still space for further improvements — with a focus on actual adoption. This however “requires a further maturity related to non-technological elements”, such as “readiness level, levels of proficiency of Operators and personnel, logistics support or other similar capabilities that are not taken into consideration in the analysis performed”. Therefore, “Further campaigns with End-Users are recommended, especially in multiple scenarios and in different locations.”
Only 2 deliverables — the Communication, Dissemination and Exploitation Final Report and the COMPASS2020 Adoption Roadmap — are available to the public some three years after project completion, providing little to no transparency on the actual outputs of COMPASS2020.
Crucially, all ethics-related deliverables are absent from public view and scrutiny at the time of writing.
We also lack foundational details about what “AI” is actually used for within the project, what kind of automated decision-making is involved, and all the technical components that were developed over the course of the project (which ones built on previous projects?).
Its relationship with other related projects — e.g., ROBORDER — should also be analyzed in much greater detail, but we unfortunately cannot given the structural opacity surrounding COMPASS2020 deliverables and the (current?) inaccessibility of its official website.
