Due to changes in the global industrial structure, the number of employees in the manufacturing industry has decreased in developed countries. One of solutions to this situation offered in Industry 4.0 is “the utilization of robots and AI as alternatives to skilled workers.” This solution has been applied to various operations conventionally performed by skilled workers and has yielded consistent results. A skilled worker has two skills, namely, “physical operation skill” and “decision making skill,” which correspond to the utilization of robots and AI, respectively.
Conventionally, robots have simply played back programs they were taught. However, owing to feedback technologies using force, position, or various other sensors, robots have come to be able to perform smart operations. In some of these, the capabilities of robots exceed those of human workers. For example, while humans are highly adaptive to various operations, it is difficult for them to maintain a constant force or position for long periods of time.
Generally, humans make decisions about operations according to their experience, and this experience is gained from many instances of trial and error. Now, the trial-and-error learning of AI has become significantly superior to that of humans in terms of both number and speed. As a result, many systems can find operational strategies or answers much faster than humans can.
This special issue features papers on robot hands, path planning, kinematics, and AI. Papers related to robot hands present an actuator using new principles, new movements, and the realization of the precise sense of the human hand. Papers related to path planning present path generation on the basis of CAD data, path generation using image processing, automatic path generation on the basis of environmental information, and the prediction of error and correction. Path generation using VR technology and error compensation using an AI technique are also presented. A paper related to kinematics presents the analysis and evaluation of a new mechanism with the aim of new applications in the field of machining.
In closing, I would like to thank the authors, reviewers, and editors, without whose hard work and earnest cooperation this issue could not have been completed and presented.
The task of screwing is based on a set of actions with no added value, requiring precision, attention, and repeatability. These set of actions could consist of alienating and demanding activity for a human operator. Collaborative robotics can facilitate the performance of such tasks. This investigation focuses on the development of a smart station for the automated screwing of fittings in pneumatic manifolds. The collaborative robot Sawyer produced by Rethink Robotics is equipped with an appropriate end-effector and was utilized to receive the fittings from a vibrating feeder towards the end-effector. This facilitated centering of the fittings on the threaded holes, and the performance of the screwing task on a set of manifolds placed on a rotating station. The design of the end-effector and its prototype is described. In addition, the proposed automated process was experimentally tested and its effectiveness was validated.
A method for extracting the machining region from a 3D CAD model in Standard Triangulated Language (STL) format and automatically generating a tool path is proposed. First, a method is proposed for extracting the machining region and obtaining the geometrical features such as a convex or concave shape from only the 3D CAD model in STL format. The STL format uses only triangular mesh data and drops all information, which is necessary for extracting the removal volume for the machining and geometrical characteristics. Furthermore, the triangular mesh size is non-uniform. A contour line model is proposed in which the product model is minutely divided on the plane along any one axial direction and is represented by points at intervals below the indicated resolution obtained from the contour line of the cross section of the product. Subsequently, a method is proposed to determine the machining conditions for each extracted machining region and automatically generate a tool path according to the geometrical features of the machining region obtained. A machining experiment was conducted to validate the effectiveness of the proposed method. As a result of the machining experiment, it was confirmed that the tool path automatically generated from the 3D CAD model in STL format can be machined without any problems and with a practical level of accuracy.
In this study, we implemented a constrained motion planner that enables robot manipulators to flip large and heavy objects without slippage while continuously holding them. Based on the soft-finger maximum friction torque, we developed a constraint relaxation method to estimate the critical rotation angle that a robot end effector can rotate while avoiding in-hand slippage. The critical rotation angle was used in a motion planner to sample safe configurations and generate slippage-free motion. The proposed planner was implemented using a 6-degree-of-freedom robot arm and a 2-finger robotic gripper with rubber pads attached to the fingertips. Experiments were performed with several objects to examine and demonstrate the performance of the planner. The results indicated satisfying planning time and the elimination of object slippage.
Disassembly is a vital step in any treatment stream of waste electrical and electronic equipment (WEEE), preventing hazardous and toxic chemicals and materials from damaging the ecosystem. However, the large variations and uncertainties in WEEE is a major limitation to the implementation of automation and robotics in this field. Therefore, the advancement of robotic and automation intelligence to be flexible in handling a variety of situations in WEEE disassembly is sought after. This paper presents an ontology-based cognitive method for generating actions for the disassembly of WEEE, with a focus on LCD monitors, handling uncertainties throughout the disassembly process. The system utilizes reasoning about relationships between a typical LCD monitor product, component features, common fastener types, and actions that the system is capable of, to determine 4 key stages of robotic disassembly: component identification, fastener identification, disassembly action generation, and identification of disassembly extent. Further uncertainties in the form of possible failure of action execution is reasoned about to provide new actions, and any unusual scenarios that result in incorrect reasoning outputs are rectified with user-demonstration as a last resort. The proposed method is trialed for the disassembly of LCD monitors and a product unknown to the system, in the form of a DVD-ROM drive.
A localization system using reflective markers and a fisheye camera with blinking infrared lights is useful and safe for mobile robot navigation in an environment with coexisting humans and robots; however, it has the problems of low robustness and a small measurable range for marker detection. A large, square-shaped reflective marker, with solid and dotted edges, is proposed for more reliable localization of indoor mobile robots. It can be easily detected using Hough transform and is robust for occlusion. The coordinates of the four corners of the square-shaped marker determine the robot’s localization. Infrared lighting with a new LED arrangement is designed for a wide measurable range via brightness simulation, including the effect of observation and reflection angles. A prototype system was developed, enabling the 2D position and orientation to be detected with an accuracy of 60 mm and 3◦, respectively, within a 4 m2 area.
This study aims to automate deburring resin workpieces as generated in injection molding by an articulated robot using sensorless shape-tracing deburring technology. Because resin workpieces largely vary in their individual shape differences, as well as installation errors, it is difficult for an articulated robot, which operates solely based on the given teaching in general use, to precisely deburr the workpieces owing to its precision deficiency. In this study, a deburring technology called “shape-tracing deburring” was developed to prevent a tool from breaking into a workpiece while absorbing any positional errors, based on a mechanism capable of mechanically maintaining the force between the tool and the workpiece constant in relation to the shape of the latter. In this way, an articulated robot can stably deburr the workpiece by following any changes in the workpiece shape. In this report, the principle and system of the shape-tracing deburring technology capable of mechanically tracing the workpiece shape without a sensor are discussed. Furthermore, the effectiveness of the developed shape-tracing deburring technology is demonstrated through an example of deburring a resin molded article with an actual cutter complete with a shape-tracing part.
This paper proposes a robot teaching method using augmented and virtual reality technologies. Robot teaching is essential for robots to accomplish several tasks in industrial production. Although there are various approaches to perform motion planning for robot manipulation, robot teaching is still required for precision and reliability. Online teaching, in which a physical robot moves in the real space to obtain the desired motion, is widely performed because of its ease and reliability. However, actual robot movements are required. In contrast, offline teaching can be accomplished entirely in the computational space, and it requires constructing the robot’s surroundings as computer graphic models. Additionally, planar displays do not provide sufficient information on 3D scenes. Our proposed method can be employed as offline teaching, but the operator can manipulate the robot intuitively using a head-mounted device and the specified controllers in the virtual 3D space. We demonstrate two approaches for robot teaching with augmented and virtual reality technologies and show some experimental results.
In this study, we evaluated the motion accuracy of a large industrial robot and its compensation method and constructed an off-line teaching operation based on three-dimensional computer aided design data. In this experiment, we used a laser tracker to measure the coordinates of the end effector of the robot. Simultaneously, the end-effector coordinates, each joint angle, the maximum current of the motors attached to each joint, and rotation speed of each joint were measured. This servo information was converted into image data as visible information. For each robot movement path, an image was created; the horizontal axis represented the movement time of the robot and the vertical axis represented the servo information. A convolutional neural network (CNN), a type of deep learning, was used to predict the positioning error with high accuracy. Subsequently, to identify the features of the positioning error, the image was divided into several analysis areas, one of which was filled with various colors and analyzed by the CNN. If the prediction accuracy of the CNN decreased, then the analysis area would be identified as a feature. Thus, the features of the Y-axis positioning error were observed for teaching each joint angle in the opposite direction just after the start of the motion, overshoot of the rotational joint current, and the change in the swivel joint current.
Robot-type machine tools are characterized by the ability to change the tool posture and machine itself with a wider motion range than conventional machine tools. The motion of the robot machine tool is realized by simultaneous multi-axis control of link mechanisms. However, when the robot machine tool performs a general milling process, some problems that affect the machining accuracy occur. Moreover, it is difficult to identify the motion errors of each axis, which influence machining accuracy. Thus, it is difficult to adjust the servo gain and alignment error. In addition, the machining performance is unidentified because of the rigidity differences when the posture changes. In this study, the focus was on robot-type machine tools consisting of a serial and a parallel link mechanism. A geometric model is described, and the forward kinematics model is derived based on the geometric model. Machining tests were then carried out to evaluate the machining accuracy by measuring the machined surfaces and the simulated motion of the tool posture based on the proposed forward kinematics model to identify the mechanism that affects the machined surface roughness and surface waviness. As a result, it was shown that the proposed model can separate and reproduce the behavior of each axis of the machine. Finally, it was clarified that the behavior of the second axis has a great influence on the tool posture and machined surface.
An on-machine measurement (OMM) system is an effective apparatus for achieving an efficient profile compensation and improving machining conditions in ultrahigh precision machining. Herein, we report a new OMM system with a confocal chromatic probe on a five-axis ultrahigh precision machine tool constructed using a real-time position capturing method. The probe and machine tool positions are captured synchronously using a personal computer to generate profile measurement data. Long- and short-term stability, micro step response, and repeatability tests using an optical flat indicates that the system has a precision of approximately ±10 nm. The profile measurement test using a reference sphere indicates that the precision of the OMM system deteriorated at a large slope angle of ±45°. However, the overall accuracy is estimated to be within ±100 nm at a slope angle within ±15°. The linearity test at various slope angles indicates that the decrease in reflected light from a mirror-like surface deteriorates the performance of the probe.
Superconductive-assisted machining (SUAM) is a polishing method that employs a magnetic levitation tool, which is based on a superconductive phenomenon called the pinning effect. Since the tool magnetically levitates, the issue of tool interference is eliminated. In this study, in order to set up the polishing conditions of the magnetic levitation tool, we evaluated the relation between the flux density distribution relative to the tool position and the holding force acting on the magnetic levitation tool to maintain its initial position, set by field cooling by the superconducting bulk. For the holding force, we measured the attractive, repulsive, restoring, and driving forces. We found that the greater the holding force, the smaller the initial distance between the superconducting bulk and the magnetic levitation tool. The attractive force was found to peak when the levitated tool was displaced 6 mm from an initial position of 9 mm from the bulk, and it became only the self-weight of the magnetic levitation tool at displacements of 30 mm and above, where the pinning effect broke down. We then evaluated the polishing characteristics for SUS304 and A1100P at a tool displacement that results in the maximum attractive force. In the polishing experiment, we employed a water-based diamond slurry because the temperature of the workpiece was close to room temperature. We found that it was possible to polish SUS304 and A1100P while avoiding the effects of magnetization due to the polishing pressure or induced currents that accompany the rotation of the metal plate. The arithmetic average roughness, Ra, of A1100P was relatively high due to the effect of scratches, while that of SUS304 improved from 92 nm before polishing to 55 nm after polishing when polished with grains with a diameter of 1 μm.
There are many changing factors in a greenhouse, and the traditional control method has been unable to obtain a good control effect. In this study, focusing on the fuzzy neural network (FNN), the principles of two control methods and the advantages of their combination were analyzed, an intelligent remote control system for a greenhouse based on the FNN that controls the temperature and humidity was designed, and a simulation experiment was performed in the Simulink environment. The results demonstrated that compared with the traditional proportion, integration, differentiation (PID) control system and the genetic algorithm + fuzzy PID control system, the FNN-based system designed in this study achieved better performance in temperature and humidity control. The temperature error of the FNN-based system was smaller than 1◦C, the humidity error was approximately 2%, and the change in the error values was stable. The experimental results verify the reliability of the FNN and provide some theoretical basis for the intelligent control of greenhouses.