Affect-based intelligent motion control for human-robot collaborative assembly in manufacturing was developed, and the effects of the dynamic affect-based control on human-robot collaboration (HRC) and assembly performance were investigated. An anthropomorphic robot with affect display ability was used to collaborate with a human in an assembly task where the human and the robot collaboratively assembled three parts. Firstly, in order to receive bioinspiration, the affective features in a human-human collaborative assembly task were studied. Secondly, based on the affective features of humans, an affect-based intelligent motion control strategy for the robot was proposed so that the robot could dynamically adjust its affective states like humans with changes in task situations during the human-robot collaborative assembly. The proposed affect-based motion control was experimentally evaluated for HRC and assembly performance, and the results were compared with that when the robot collaborated with its human counterpart with no affect and a static affect displays. The results showed that the static affect produced better HRC and assembly performance than that the robot produced with no affect. However, the dynamic affect produced significantly better HRC and assembly performance than that the static and the no affect displays produced. The results encourage to employ anthropomorphic robots with dynamic affect-based motion control strategies to collaborate with humans in manufacturing to improve HRC and manufacturing performance.

In the first step, a 1-DOF power assist robotic system (PARS) was developed for object manipulation. The dynamics for human-robot co-manipulation of objects was derived that included human's weight perception. An admittance control scheme with position feedback and velocity controller was derived using the weight-perception-based dynamics. Human subjects lifted objects with the PARS, and human-robot interactions (HRI) and system characteristics were analyzed. HRI was expressed in terms of physical HRI (maneuverability, motion, safety, stability, and naturalness) and cognitive HRI (trust and workload), and performance was expressed in terms of manipulation efficiency and precision. A constrained optimization algorithm was used to determine the optimum HRI and performance. Results showed that inclusion of weight perception in the dynamics and control was effective to produce optimum HRI and performance for a set of hard constraints. In the second step, a novel variable admittance control algorithm was proposed, which enhanced the physical HRI, trust, precision and efficiency by 34.63%, 31.86%, 3.85% and 4.92% respectively, and reduced cognitive workload by 35.36%, and helped achieve optimum HRI and performance for a set of soft constraints. Effectiveness of the control algorithm was justified using a multi-DOF PARS for manipulating heavy objects.

This paper presents the analysis of electro-mechanical design, materials, dynamics and control of a novel variable impedance compact compliant series elastic actuator (SEA) to improve human-robot interactions in human-friendly soft robotics applications. The design consists of a servomotor, a ball screw, a torsional spring connecting the servomotor and the ball screw via a pair of spur gear, and a set of translational springs connecting the ball screw nut to the output link. The translational springs have low stiffness and these are used to handle low force operations that reduce non-linear friction, output impedance, impact etc. The torsional spring is in the high speed range, has high effective stiffness and it enhances the system bandwidth for large force operations when the translational springs are fully compressed. Design, materials for construction, kinematics and working principle of the actuator are analyzed. Then, dynamics modeling and control on the physical implementation of the actuator for different conditions are analyzed. The conditions are: (i) open loop transfer function with load end fixed, (ii) closed loop transfer function with load end fixed, and (iii) output impedance with load end free for (i) low force, and (ii) high force cases. Then, the advantages of the design over its existing counterparts are discussed and its potential applications are mentioned. This novel SEA overcomes the major limitations of the existing SEAs, and can be used to develop human-friendly soft robots for various applications.

In this paper, we present the design and evaluation of a novel control method for a power assist robotic system for manipulating objects with it by human subjects. We consider the manipulation of objects in harmonic motion (object is lifted up and lowered down repeatedly) as we think that this motion may be very practical and useful in object manipulation. We develop a 1 DOF power assist system. The subjects manipulate different sizes of objects with the system in harmonic motion. We analyze human characteristics such as weight perception, load force, object motions etc. We find that the load force and acceleration are excessive that reduce the performances of the system. We then optimize the perceived heaviness. We then design and implement a novel control method based on human characteristics to reduce load force and acceleration. Results show that the novel control reduces load force and acceleration, optimizes perceived heaviness and thus makes the performances (maneuverability, safety etc.) satisfactory. We also compare this novel control designed for object manipulation in harmonic motion to another novel control designed for lifting objects with power-assist in linear vertical motion. Finally, we propose to use the findings to develop control method for power assist devices for manipulating heavy objects in industries that may enhance interactions between humans and robots.

A power assist robot reduces perceived weight of objects lifted with it. However, root cause of reduced heaviness and factors affecting the heaviness are still unclear though the knowledge on the root cause/factors could be used to modulate human's feelings and the interactions between human and robot. This paper investigated the causes/factors behind the reduced heaviness. We adopted two strategies for this. Firstly, we included weight perception in robot dynamics (and control) in such a way that the mass parameter of the inertial force was considered different from that of the gravitational force because the perceived weight is different from the actual weight. Secondly, we compared weight perception, load forces and motions for power-assisted manipulation to that for manual manipulation and identified some clues regarding the causes/factors of reduced heaviness. We found that perceived weight, load force and its rate, velocity and acceleration for power-assisted objects were lower than that for manually lifted objects. We observed time delays in position sensing, force sensing, servomotor etc. for the power-assisted objects, however, the delays were almost absent for the manually lifted objects. We argued that the delays were the root cause of reduced heaviness and the elements of the system contributing to the delays were the factors affecting the perceived heaviness. To cross-check it, we increased time constant of the servomotor and found that the increased time constant reduced perceived weight. The findings may be used to develop power assist robots for manipulating heavy objects in industries that may help improve/modulate interactions between robots and users.

We hypothesized that weight perception (WP) due to inertia might be different from WP due to gravity for lifting an object with a power assist robot (PAR). Objects were lifted with a PAR independently under three different lifting schemes-unimanual, bimanual, and cooperative lift. Then, psychophysical relationships between actual and power-assisted weights (PAWs) as well as excess in load forces (LFs) were determined for each scheme separately. A novel control strategy was introduced to reduce the excess in LFs for each scheme. Finally, we proposed to use the findings as design guidelines to design PARs for lifting heavy objects in industries that would improve HRI in terms of human, robot and system.

We developed a 1 DOF power assist robot for lifting objects. We 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 robot. We established psychophysical relationships between the actual weights and the power-assisted weights for the objects lifted with the robot, and also determined the excess in load forces that the subjects applied for three independent lifting schemes or grasp configurations: (i) unimanual lift, (ii) bimanual lift, and (iii) cooperative lift. We also compared the weight perceptual and load force features for the unimanual lifts to that for the bimanual and cooperative lifts. We then modified the power-assist control using a novel control strategy based on the weight perceptual and load force features. The control modification reduced the excessive load forces applied by the subjects in each lifting scheme and thus enhanced maneuverability, naturalness, ease of use, stability, safety etc. of the robot system significantly. Finally, we proposed using the findings to design human-friendly power assist robots for carrying heavy objects in various industries.

An object lifted with a power assist robot is always perceived lighter than its actual weight. But, the human operator cannot differentiate between the power assisted weight and the actual weight and eventually applies load force (vertical lifting force) according to the actual weight of the object. This faulty force programming (excessive load force) gives faulty motions to the power assist robot and jeopardizes its operability, maneuverability, ease of use, naturalness, human-friendliness, safety etc. In this paper we assume that these problems still exist with the power assist robots because human's weight perception is not included in the design and control of the conventional power assist robots .We hypothesize that human's perception of weight due to inertial force may be different from the perceived weight due to gravitational force for lifting object with a power assist robot. Based on this hypothesis, we designed a 1 DOF power assist robot and established a psychophysical relationship between the actual weights and the power assisted weights for the objects lifted with the robot. We also determined the excess of the load forces that humans applied. Then, we modified the control system of the power assist robot based on the psychophysical relationship and the load force characteristics. The modification of the control system reduced the peak load forces applied by humans and thus enhanced maneuverability, naturalness, ease of use, stability, safety etc. of the robot system significantly. Finally, we proposed using the findings to design human-friendly power assist robots for carrying heavy objects in various industries.

In this paper, we designed a 1DOF power assist system for lifting objects. Human's vertical lifting force, comprises of inertial force and gravitational force, was considered as the desired dynamics of the system. We hypothesized that human's perception of object weight due to inertial force might be different from the perceived weight due to gravitational force for lifting object with the power assist system. The hypothesis meant that the mass parameter of the inertial force might be different from the mass parameter of the gravitational force. We then designed the control system of the power assist system based on the desired dynamics and simulated the system using MATLAB. We then psychophysically determined the optimum mass parameters for the inertial and the gravitational force components of the desired dynamics of the power assist system on the basis of human's perception of object weight. The results showed that human's weight perceptual consideration with the dynamics (control system) of the power assist system enhanced maneuverability, stability, naturalness, ease of use, safety etc. when lifting objects with the system. Finally, we suggested using the findings to design human-friendly power assist systems for manipulating heavy objects in various industries.