Design and Control of a Power Assist Robot System for Lifting Objects Based on Human's Weight Perception

三重大学大学院工学研究科博士後期課程システム工学専攻Power assist robot systems assist the humans perform tasks by augmenting humans’ abilities and skills. However, the research on power assist robots is now confined to a few fields of applications such as healthcare, rehabilitation, old care etc. We think that manipulating heavy objects in industries is another potential field of application of the power assist devices. Manual manipulation of heavy objects is very tedious, causes health related problems to humans and restricts work efficiency. On the contrary, autonomous robotic systems usually do not provide required flexibility in object manipulation. Hence, I argue that suitable power assist robots may be conveniently and efficiently used for handling heavy objects in industries such as agriculture, forestry, mining, construction, manufacturing and assembly, transport and logistics, military activities, disaster and rescue operations, meat processing etc. However, such power assist robots are not available in practices in industries. On the other hand, a power assist robot reduces the weight of the object manipulated with it, and hence the forces required to manipulate the object with power-assist should be lower than that required to manipulate the object manually. However, limitations with the conventional power assist devices are that the human user cannot correctly perceive the weight of the object before lifting it with the robot and eventually applies excessive load force (manipulative force tangential to grip surfaces). The excessive load force results in sudden increase in acceleration, fearfulness of the user, lack of stability and maneuverability, injuries, fatal accidents etc. I assume that suitable power assist robots for manipulating heavy objects are not found in industries and their interactions with users are unsatisfactory because (i) specialized power assist robots for handling heavy objects have not been developed yet, and (ii) the conventional power-assist controls do not consider human characteristics such as weight perception, load forces and object’s motion features. This thesis deals with the design and control of a power assist robot system for lifting objects. A one-degree-of-freedom (vertical up-down motion) power assist system for lifting objects was developed. The system was actuated by an AC servo motor. It was hypothesized that human’s perception of weight due to inertial force might be different from the perceived weight due to gravitational force for lifting an object with a power assist system. The system was simulated for different sizes of objects (sizes of the objects were between 5x6x9cm and 5x6x16cm, and actual weights were between 0.5kg and 1.5kg) and a series of experiments (simulation) were conducted to derive various results as the following : (i) Optimum maneuverability conditions for lifting objects with the power assist systemwere determined. (ii) Psychophysical relationships between perceived weights and actual weights for the objects lifted with the system were determined. Psychophysical studies show that the power assist system reduces the weights of the lifted objects to 40%, 30% and 25% of the actual weights for unimanual, bimanual and cooperative lifting tasks respectively. The results were then validated (cross-checked) by a comparison study between constrained and unconstrained lifting of objects. The excess in the load forces were also determined. The studies show that human operators apply 6.38, 7.30 and 8.0 times larger than the actually required load forces for unimanual, bimanual and cooperative lifting tasks respectively. (iii) A novel control strategy based on human's weight perception and load force features was implemented that reduced the excessive load forces and thus enhanced maneuverability, safety, naturalness etc. for lifting objects with the system in unimanual, bimanual and cooperative lifting tasks. (iv) Weight perception and load force characteristics for power-assisted manipulation in harmonic motion (objects were lifted up and lowered down repeatedly) were also studied. The results show that perceived weights and load forces for the objects manipulated in harmonic motion are less than that for the objects manipulated in linear vertical motion (lifting objects up only). The effectiveness of the novel control for harmonic motion was also justified. (v) Effects of friction between lifted object and operator's hand on weight perception and load force characteristics were also studied. The results show that increase in friction between object and operator's hand reduces perceived weight and load force. (vi) Operator’s weight perception and load force characteristics in some worst-cases(uncertain, doubtful etc.) were studied. The results show that the system performances reduce at the worst-cases. Then the effectiveness of the novel control was justified for the worst-cases and an advanced control was proposed for the worst-cases. (vii) The root causes of reduced heaviness of objects lifted with the power assist system were identified that can be used to modulate human’s perceived heaviness. It was found that the time delays in position sensing, force sensing, servomotor etc. are the root causes of the reduced heaviness. Relationships between mechanical time constant of the servomotor and human's heaviness perception were also investigated. Results show that increase in mechanical time constant reduces perceived heaviness and load forces. The results also show the techniques for simultaneous optimization of motion and perceived heaviness. (viii) The feasibility of lifting objects with the power assist system at zero-gravity and zero-inertia was studied and found that zero-gravity and zero-inertia were not feasible. (ix) Human’s weight perception and load force features for power-assisted manipulation were compared to that for manual manipulation to support and cross-check the results. The findings as a whole were compared to that derived in similar works available in published literature etc. Finally, it was proposed to use the findings to develop power assist robot systems for manipulating heavy objects in industries such as agriculture, forestry, mining, construction, manufacturing and assembly, transport and logistics, military activities, disaster and rescue operations, meat processing etc. that may enhance interactions with human users in terms of maneuverability, safety, ease of use, stability, naturalness etc