<p>Affective robots are of interest for many applications, to accompany people, assist in education, and reassure patients. However, the principles for control of such a robot are not yet established, and its specifications are not clear. Here we propose a framework for a lively robot that can reason to make decisions and maintain emotions and intentions. The Affective RObotic Mind (AROM) model is designed and implemented, aiming to enhance the interpretability of affective robot behaviour for humans in various social situations encountered by the robot. Inspired by biological examples, AROM integrates robot decision-making and behaviour-generation modules based on consistent affective support for taking action based on decisions and expressing internal states via body movements. The core factors that adjust decision-making and movement patterns are abstracted from the physiological system that is shared by most vertebrates. The system is thus intended as an exemplary architecture for biomimetic affective robot design. To evaluate the achieved affective behaviour in the robot, experiments on human subjects are conducted related to the function of the decision-making module and the emotion-related enhancement of interpretability of the behaviour-generation module. The project exemplifies the potential of biomimetic design of robot emotion expressions and its promising value in shaping robots’ personality and behaviour logic.</p>