First generator of missions for robot swarms. Creates XML configuration files (.argos) for the ARGoS3 simulator.
For more info about the generator, see: A. Ligot, A. Cotorruelo, E. Garone, and M. Birattari (2022). Towards an empirical practice in off-line fully-automatic design of robot swarms. IEEE Transactions on Evolutionary Computation, 26(6):1236-1245.
The generator of strings of parameters resembles the one used in IRACE. It takes a file of parameters description as input. These parameters description will include the range of values that can be taken, and conditions (logical expressions) to create relations between parameters. The generator samples the values for the parameters that meet the conditions.
Example: mission --m categorical ('foraging', 'agg', 'dg') NA expLength --el ordinal (60, 90, 120, 150, 180) NA robots --r integer (15, 25) NA initPosit --ip categorical ('area', 'fixed') NA initOrient --io categorical ('uniform', 'fixed') NA floorCol --fc categorical ('white', 'black', 'gray') NA
Currently, for the reference model RM1.1/2, all the missions that we liked can be categorized into three classes: foraging, aggregation, and directional gate.
Conditions are a bit more complex that the ones of IRACE. For example, in a foraging mission, at least two patches of different colors need to be instantiated in the environment: one for the nest, one for the food source. Also, if obstacles are instantiated, they should allow for a clear path between the nest and the food source. The generator incorporates these predefined conditions, maybe in classes dedicated to the missions.
Example: nPatchesFor --npf integer (2, 3) mission == 'foraging' nPatchesAgg --npa integer (0, 3) mission == 'agg' nPatchesDg --npdg integer ? mission == 'dg'
typePatchFor1 --tpf1 categorical ('circ', 'rect') nPatchesFor > 0 dimPatchFor1 --dpf1 real (0.2, 0.6) nPatchesFor > 0 posPatchFor1 --ppf1 categorical {'unif'} nPatchesFor > 0 rangeXPatchFor1 --rxpf1 real (-1.0, 1.0) posPatchFor1 == 'unif' rangeYPatchFor1 --rypf1 real (-1.0, 1.0) posPatchFor1 == 'unif' centerXPatchFor1 --cxpf1 real function() randomly sample X and Y + checkOverlap centerYPatchFor1 --cypf1 real function() colorPatchFor1 --cpf1 categorical {'w', 'g', 'b'} - floorCol nPatchesFor > 0
typePatchFor2 --tpf2 categorical ('circ', 'rect') nPatchesFor > 1 dimPatchFor2 --dpf2 real (0.2, 0.6) nPatchesFor > 1 posPatchFor2 --ppf2 categorical {'unif', 'relation'} nPatchesFor > 1 relPatchFor2 --rpf2 categorical {1} #index of patch to be in relation with. rangeXPatchFor2 --rxpf2 real (-1.0, 1.0) posPatchFor2 == 'unif' rangeYPatchFor2 --rypf2 real (-1.0, 1.0) posPatchFor2 == 'unif' distPatchFor2 --dpf2 real (0.0, 0.6) posPatchFor2 == 'relation' centerXPatchFor2 --cxpf2 real function() if posPatchFor2 == 'unif', randomly sample X and Y + checkOverlap posPatchFor2 == 'relation', sample X and Y so that + checkOverlap centerYPatchFor2 --cypf2 real function() colorPatchFor2 --cpf2 categorical {'w', 'g', 'b'} - floorCol nPatchesFor > 1 AND if nPatchesFor==2, colorPatchFor2 != colorPatchFor1
nObstacles --no integer (0, 5) NA obstacle0 --o0 real (0.05, 0.80) nObstacles > 0 distObs0 --dist0 categorical ('random', 'side') nObstacles > 0, side iff at least one rectangular patch exists