Abstract This dissertation develops the intelligent-Coalition Formation framework for Humans and Robots (iCiFHaR), an intelligent decision making framework for multi-agent coalition formation. iCiFHaR is a fifirst of its kind that incorporates a library of coalition formation algorithms; employs unsupervised learning to mine crucial patterns among these algorithms; and leverages probabilistic reasoning to derive the most appropriate algorithm(s) to apply in accordance with multiple mission criteria. The dissertation also contributes to the state-of-the-art in swarm intelligence by addressing the search stagnation limitation of existing ant colony optimization algorithms (ACO) by integrating the simulated annealing mechanism. The experimental results demonstrate that the presented hybrid ACO algorithms signifificantly outperformed the best existing ACO approaches, when applied to three NP-complete optimization problems (e.g., traveling salesman problem, maximal clique problem, multi-agent coalition formation problem).