Robotics technology is approaching a state where highly capable autonomous platforms can directly interact with human companions. This paper explores the ability of a human and manipulator to efficiently complete independent tasks in a shared physical workspace without explicit communication of their planned actions. With focus on quantifying the impact of robot presence on human task performance and workload, a series of human-subject experiments were conducted in which a seated human was

more »

... to perform cognitive and mobility tasks in this shared workspace. The manipulator could unintentionally interrupt these activities, blocking the field of view or physical path to a worksite while completing its own tasks. To avert impending conflict, the robot received and reacted to ideal sensor data identifying the human's tasks upon initiation. Tests over a progressively complex set of tasks for the human-manipulator team were conducted. Results indicate minimal impact on a human companion's productivity when a manipulator is introduced to the workspace, provided it responds quickly to avoid blocking constraints. Future work involves maturation of the robot's perception and prediction capabilities followed by additional shared workspace experiments. Nomenclature A, B; : : : = types of scenarios a i−1 = distance fromẐ i−1 toẐ i measured alongX i−1 , in. d i = distance fromX i−1 toX i measured alongẐ i , in. N = number of moves per second for joint angle waypoint calculations R i = Task Load Index adjusted rating for an single load source i R W = Task Load Index weighted rating, or overall workload r i = Task Load Index rating T s = start time, when at initial pose T f = arrival time, when at final pose W i = Task Load Index weight α i−1 = angle fromẐ i−1 toẐ i measured aboutX i−1 , deg θ i = (variable) angle fromX i−1 toX i measured aboutẐ i , deg θ si = starting angle of initial pose for joint angle i θ fi = resulting angle of final pose for joint angle i