Distributed controller architectures have been proposed for Software-Defined Networking (SDN) to ensure scalability and reliability. One major drawback of existing architectures is regarding the uneven load distribution among controllers stemming from the static controller-switch binding. Though dynamic binding has been proposed to alleviate this issue by re-associating switches from overloaded controllers with underutilized ones, this process adds a considerable amount of complexity to the system and may incur significant network delay. To address these issues, three strategies will be explored in this proposed research to achieve effective and efficient load balancing and resource scheduling in distributed SDN controller architectures. First, a new bindingless architecture for distributed controllers will be proposed to achieve load balancing without requiring time-consuming switch migration. Second, evolutionary reinforcement learning techniques will be investigated to dynamically learn high-performing scheduling policies, enabling adaptive and efficient requests dispatching and controller scheduling. We expect that the proposed method can be effectively applied to a wide range of real world scheduling problems in dynamic environment. Third, a utility-based evolutionary computation technique will be developed to effectively decide the controller placement in the network. It systematically balances the trade-off in between expected improvement in network performance and potential risks of performance degradation. This technique can be widely adapted to resource allocation problems and presents a systematic way to handle uncertainties.