111 5 IMPLEMENTATION CONCEPT system which allows us to explain it to ourself or other people and investigate it before it is constructed. Secondly, it dramatically increases the understanding of the modelled system which is often more beneficial than the description or analysis results themselves. Both reasons are our motives to model the implementation concept using CPNs. We used CPN Tools to create a simulation that allows us to check the monitoring and verification process of the observation stage of the Glass Box framework and run a simulation-based performance analysis (Jensen, Kristensen, & Wells, 2007). CPN Tools has several settings to conduct an automatic simulation based on a random number generator to calculate the effects of occurring steps (Jensen, 1994). We created a simulation that shows the steps of a pre-flight mission planning and post-flight mission evaluation process for autonomous surveillance drones which is not too complex as an example. We based the processes on the scenario described in section 1.2. As reference we used information obtained in several conversations with domain experts in the Dutch Ministry of Defense and the JFCOM-UASPocketGuide-the US Army Unmanned Aerial Systems manual (JUAS-COE, 2010). The CPNs are uploaded as Supplementary Materials (https://github.com/responsible-ai/DroneCPN). Pre-Flight Mission Planning Process In the pre-flight mission planning process, first the steps are modelled to check the prerequisites for a mission; i.e., the availability of a map and the status of the weather conditions (see Figure 18). Next the compliance criteria Area of Operation, Restricted Operating Zone, and Electronic Warfare Threat are checked and if these are complied with, the flight path is calculated. If, for example, the boundaries of the Area of Operation are not known and this criterion is not complied with, then the process enters a feedback loop in which the boundaries of the Area of Operation are requested. When all criteria are met the approval process is triggered and sequentially a drone is requested. In the case that the mission is not approved, the reason for disapproval needs to be solved first in order to continue the process. The pre-flight mission planning process is modelled with several feedback loops. For example, if there is no map available then a map is requested or if the weather conditions are adverse than the mission is replanned (Figure 18). In the final step the mission is flown and, upon completion of all the steps, the preflight mission planning process ends and the drone is returned to the pool of drones and can be deployed for a next mission (see Figure 19 and Figure 20). Post-Flight Mission Evaluation Process The evaluation of the mission will be done manually and starts with two concurrent steps. The check of (1) the compliance criteria and (2) the flight path. The same compliance criteria as in the pre-flight mission planning process are checked (see Figure 21); Area of Operation, Restricted Operating Zone and Electronic Warfare Threat. If the criteria,
RkJQdWJsaXNoZXIy MjY0ODMw