Navigation is the core of mobile robot applications, but traditional configurations have great difficulties in dealing with dynamic human factors. This means that new service robots must not only undertake the task of autonomous navigation, but also be good at social interaction and consider harmonious coexistence with others. This paper designs a social navigation based on improving the comfort of human–robot interaction. The social space costs and constraints are modeled using asymmetric Cauchy functions, and predictions are made using human–human or human–robot interaction, and pedestrian encounters are considered. The difference in the degree of attention paid to oneself, front, rear, left, and right when encountering obstacles or pedestrians establishes the benchmark for the corresponding model. On this basis, a map cost function is constructed, which can use different constraints on the path and specify that the robot does not enter certain spaces, or enter specific spaces under certain circumstances. The A* and jump algorithms were modified based on the map cost function, and experiments were conducted in MATLAB. The experimental results show that the designed social comfort navigation can effectively realize the function, pedestrians’ personal space is guaranteed, and goal-oriented intentionality is understood by the robot. Understanding, coexistence and adaptability of mobile service robots are significantly improved.