Adjustable autonomy architecture provides various ways for a human operator to participate as a member of a human-robot team in improving the performance of the team by resolving issues that the robots cannot deal with or performing tasks that the robots alone would unable to do. According to the level of involvement of the human operator, the robots have to adjust their level of autonomy and, in consequence, the operation mode of the overall system shifts. This paper deals with the implementation issues of seamless switching when the level of autonomy of the human-robot team shifts from one level to another. Especially, we focus on developing reliable methods for monitoring the task progress and maximizing the system flexibility by coping with the detailed differences between humans and robots in their characteristics of motions and their choices of positions, paths, and sequences of sub-goals to achieve a given task. To test and motivate the proposed methods, we have assembled three heterogeneous robots which work together to dock both ends of a suspended beam into stanchions.