Ropero et al. Figure 10. Demonstration of ARIES for the described Nominal scenario. Figure 10(a) shows an instance with six target points (TP1-6) distributed around the Mars surface. The UGV carries the UAV through secure locations (SL1-4 in green color) to allow the UAV to reach the TPs (blue color) without running out of energy. Figure 11. Partial PDDL plan to reach TP1 and TP2 from SL1 in the Nominal scenario. The start and duration parameters are represented in seconds. The attributes X; Y; Z follow the format Name_Value. shows the part of the plan for reaching TP1 and TP2 from the secure location SL1 [see Figure 10(b)]. The generated PDDL plan is forwarded to the Planner Reactor. Then, the Planner Reactor is in charge of translating the plan actions into a timeline-based plan (tokens) using the PDDL-Lib (see Section "FROM TEMPORAL ACTION-BASED TO TIMELINE-BASED PLANNING"). The computational cost of the translation process of the current PDDL-lib version is negligible with respect to the controller execution. But we can ensure that the performance of future versions will be tightly associated with the complexity of the world knowledge representation, i.e., to the complexity of the temporal synchronization rules which defines the interconnection among the actions of the domain. Figure 12 shows the MARCH 2019 translation of the action plan into a timeline-based plan. This timeline-based plan contains the PDDL actions (A1-11) plus the initial states (green boxes) modeled in the PDDL problem and the effect tokens (gray boxes) stated in the effects configuration file. In this way, each state variable has a complete temporal evolution from the beginning to the end of the time horizon. Now, the new plan can be executed by an architecture which follows the timeline-based paradigm such as a T-REX-based architecture. Once the plan is translated, the Planner Reactor starts the execution token by token. Figure 13 shows the execution flow for achieving TP1 and TP2 targets. First, we observe that every planned token has its own effect token, e.g., the TakingPicture token has the Idle consequence. IEEE A&E SYSTEMS MAGAZINE 51