The paper for this session was “WeBuild: Automatically Distributing Assembly Tasks Among Collocated Workers to Improve Coordination,” a collaborative work between AutoDesk Research and UC San Diego from Fraser et al. The paper concerns itself with the design of a system to facilitate assembly tasks that require multiple workers. Though the potential applications for such a technology are far reaching, the types of tasks at which it currently excels (and on which it was tested) are along the lines of IKEA or Lego assembly projects. Taking inspiration from how groups typically accomplish these tasks, the researchers designed a system that would better facilitate projects of this nature. The two major parts of the system are the central dashboard and the individual displays (mobile phones) for each worker. The central dashboard displays the overall plan for the task, along with a progress report and information on each user (e.g., what tools they possess and their current task). Each participant also has a mobile phone with specific instructions for their specific subtask. The manner in which these subtasks are assigned is the other major portion of this system. Interestingly, the authors’ algorithm works on the fly, thus making it adaptable to changes in number of participants. There are several factors that go into this assignment process, including familiarity, speed, and repetition.
The results of testing this system were, at least in my opinion, a bit weak. Though the group using the WeBuild system completed the task faster, the difference was not statistically significant. The users of WeBuild, however, did spend a larger proportion of their time working. Their activity was also far more parallel than the control group. This perhaps suggests that WeBuild enables more efficient work. Another interesting result was that WeBuild users spent more of their working time in silence, as well as rating themselves as less aware than the control group.
There are two major questions I have after reading this paper. First, how adaptable is this system to other, more complicated assembly tasks with a much larger pool of participants? The authors address this briefly at the end of the paper, but they seem to rest the burden on the task model. I’m curious if there will be a point at which the effort required to completely model a task for WeBuild will overshadow the benefits gained from using the system. The other question I have about the system is about its use in creative tasks. As the system seems, based on the data collected, to dampen intra-group communication and overall task awareness, I would be curious to see how this affects the creative process. Can this type of task division work to enable creativity, or is an awareness of the system as a whole necessary?