Worst-cases prediction by human in lifting objects with a power assist robot system: Effectiveness of a novel control strategy to improve the system performances in worst-cases

We constructed a 1 DOF power assist robot for lifting objects of different sizes. We hypothesized that human's perception of weight due to inertia might be different from the perceived weight due to gravity when lifting an object with the power assist robot. In this article, we particularly looked at human's load force features, weight perception and object's motions in lifting objects with the power assist robot in worst-cases situations. We called it a worst-case when the human faced any uncertainty, sudden change in work environment, doubt or unusual situation prior to or at the moment of lifting. We considered two potential worst-cases. In the first case, subject's vision was obstructed by a screen prior to lifting the object with the robot. In the second case, the object was tilted at the moment of lifting. We then critically analyzed human's load forces, weight perception and object's motions for two cases separately. We then applied a novel control technique to two cases separately to reduce the excessive load forces and to improve the system performances. We also compared the findings derived in worst-cases to that derived in usual cases (i.e., when vision was not obstructed and objects were not tilted). Finally, we proposed to use the human features and the control technique to develop human-friendly power assist robots for lifting heavy objects in industries such as manufacturing, mining, transport, construction etc.