| For each agent and episode do | | | | IF is the critical section | | Initialise , | | ELSE | | Initialise , , | | For each control step do (Loop I) | | (i) get detected data from each cell : , , , , | | (ii) get state through (8) and (9) | | (iii) get action by -greedy policy (10) | | (iv) get , through (6) and do , | | IF is the critical section | | update , through (2) and (12) | | ELSE update , , through (4) and (13) | | IF and end the algorithm | | ELSE get , , by (7) and do , , , and start loop II | | For each planning step do (Loop II) | | (i) generate flow rates for each cell : through (5) | | (ii) get the state | | (iii) get , and do , | | (iv) get action by -greedy policy | | IF is the critical section | | update , | | ELSE update , , | | IF or ( and ) go back to loop I | | ELSE repeat loop II | | EndFor | | EndFor | | EndFor |
|
