This study develops a dialogue model and a command interpretation technique for service robots to obtain interactively the necessary information of actions for the tasks requested from users. A request from users sometimes misses some data necessary to complete the requested task because a voice command is usually requested depending on context and situation. The developed command interpretation technique first divides a command into words. Then, it analyzes the content of the command paying attention to the verb used and object included. If there are some missing data in the command, it sequentially generates queries to obtain them. The applicability of the dialogue model and the technique is demonstrated by dialogue experiments using a robot arm system.