[Context and motivation] Self-adaptive robotic systems often operate in complex problem domains where mission plans are restricted by multiple requirements. These requirements often lead to context-dependent conflicts that hinder practitioners from automatically generating plans that adhere to their requirements. [Question/problem] Currently, practitioners usually solve such conflicts manually since available techniques, e.g., prioritization and requirements relaxation, are not always effective given that they do not consider the context of a conflict. Additionally, these techniques do not take the stakeholders’ varying needs into account. [Principal ideas/results] Our study applies a design science approach to identify the challenges practitioners face when handling conflicting requirements and elicit potential solutions. We develop an adaptive conflict resolution approach that iteratively combines the two different resolution strategies relaxation and prioritization, supported by human feedback. [Contribution] Our results show that practitioners commonly are confronted with complex combinations of requirements and resulting conflicts for which conventional resolution techniques reach their limits. Our approach to adaptive conflict resolution addresses these issues and was evaluated in interviews with five practitioners. The results show that interviewees deem our approach feasible for solving common conflicts in robotics and consider it more effective than manual approaches. In the future, we aim to include explanations to help humans make decisions and understand conflict resolution strategies, especially in situations shaped by complex problem domains and operating contexts.