Modelling and simulation of human path findingthrough terrain while avoiding detection
Carl Pettersson · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2013
Training military commanders using simulated battlefields currently requires a large staff of operators behind the scenes due to insufficient autonomy in the simulated agents. If a more autonomous behaviour can be achieved, simulations can be scaled up without requiring a corresponding increase in staff and associated equipment. When greater autonomy is achieved, one behavioural aspect that is interesting to develop is the agent’s movement through and strategic usage of the terrain. The aim of this project was to examine the feasibility of modelling this type of behaviour. To this end, agents were created using the cognitive framework library Soar. These could interact with a pathfinder module using the D* dynamic pathfinding algorithm. The simulation framework Ternion FLAMES ® was used to simulate the model’s interaction with the environment. To limit the complexity, the model tried to avoid all confrontation with hostile agents. The results from the simulations showed that the chosen method is capable of simulating a credible behaviour, even though much work remains in fine tuning and extending the model.