Agent-based methodologies facilitate complicated hypotheses verification, modeling and dynamic simulation. Hence, there are many interests in the geospatial information science to model rational autonomous agents due to their closer-to-realism decision-making and influence on environment. In this study, we consider the planning and task distribution among vehicle entities in the geospatial domains based on an agent-based approach. We show at the beginning of the paper, the rational agents have tendency to change their strategies to reach the highest possible satisfaction/utility. To obtain such a utility in the group of collaborative agents, equilibrium, a key concept taken from the Game theory, appears more efficient than the common multi-objective optimization. We challenge this issue according to the dependency or contention of moving agents' payoffs. Because of high complexity of determining equilibrium, i.e. exponential, an efficient non-deterministic heuristic algorithm is proposed. We get our inspiration from evolutionary computations to introduce this novel algorithm. We have evaluated our approach with several datasets and received perfectly acceptable convergence, accuracy and speed. In comparison to a pure deterministic method, retrieving the equilibrium and the optimality of the best equilibrium solution were experimented at least as 80% and 92% respectively.