In recent years, human collaboration robots (HCRs) have become an attractive proposition. In this paper, we present a simplified tactile and proximity sensor system for HCRs. The proposed proximity and tactile sensor uses a self-capacitance measurement, and it consists of two electrodes (E1 and E2), and an elastic body. The self-capacitance between either E1 or E2 and ground is measured by switching between E1 and E2. We placed four sensors on the same I2C bus line to cover the robot surface. The proposed sensor on the robot surface can detect the object before contact. In addition, the sensor can detect the pressure and position, and can discriminate the materials upon contact. The developed sensor may thus be applicable for use as a tactile sensor for HCRs.
A laser microphone that uses the self-coupling effect of a semiconductor laser is an optical microphone. It has excellent characteristics, such as a wide and flat frequency characteristic. However, as a laser microphone has a large noise component in its output, the signal-to-noise ratio is low. Therefore, the sensitivity of a laser microphone is not sufficiently high for practical use. In this work, two methods to improve the sensitivity of a laser microphone are proposed. The first method uses a sound collector that collects widespread sound waves on a laser beam. The second method uses a right-angle prism that extends the light path. By using a sound collector, the sensitivity can increase up to nine times and the signal-to-noise ratio increases up to six times. However, the sensitivity decreases for frequencies below 10 kHz, and the directivity has a sharp peak. By using a right-angle prism, the sensitivity increases up to three times and the signal-to-noise ratio increases up to six times. In addition, the sensitivity can be improved while maintaining the wide and flat frequency characteristics and the wide directivity.
Tracking a dynamic odor plume to the odor source in a real world environment using a chemical sensor is not a trivial task. In general, odor distribution in the real world is temporally and spatially fluctuated in a random manner owing to turbulent airflow. Therefore, evaluating a tracking strategy in a real world environment presents a certain difficulty in setting up a consistent environment for repetitive testing. This paper presents a simulation framework for the development and evaluation of a tracking strategy in a consistent virtual environment. A dynamic turbulent environment model with an odor source was modeled using Computational Fluid Dynamics (CFD) software, and the resulting data were used for a plume tracking simulation. This paper addresses the problem of estimating the trend of a fluctuating odor concentration for odor plume tracking. An estimation method for the odor concentration trend based on a regression was compared with the one based on an adaptive threshold. Each method was embedded into a chemotaxis-anemotaxis based plume tracking strategy to localize the source. The performances of the tracking strategy employing these different methods were compared. In addition, data smoothing was also examined to improve the effectiveness of the estimation methods. The investigation shows that the tracking strategy employing the combined smoothing-regression trend estimation has the highest success rate and minimum average time performance compared to the other tested methods. This paper also shows that selection of the key parameter of the method affects the tracking performance.
A thin-film magneto-impedance (MI) sensor has the unique property of a step-like magneto-impedance change, in the case where the sensor has an in-plane uniaxial inclined easy axis. A domain observation shows that the step-like change is due to a magnetization transition within three states, including the longitudinal single domain with parallel state, that with anti-parallel state and the inclined Landau-Lifshitz domain (ILLD). In a condition where the sensor has an easy axis of 70º relative to the short-side axis of the rectangular element, the transition is limited between the parallel and anti-parallel states despite the existence of a stable ILLD. Previous work showed that the application of a normal field could produce the ILLD state. This paper reports a control method for the magnetic domain by controlling the normal field. It is shown that the existence of a distributed canted angle relative to the normal direction in the normal field enhances the reconstruction of the ILLD domain state.
Previously, we proposed a method for driving a piezoelectric actuator using current pulses, for high-resolution and small residual vibrations. In the proposed approach, the driving pulse patterns were designed using a genetic algorithm. In this paper, positioning performance of a stage with two degrees of freedom was compared with the voltage drive by an input shaper and PI control. To enable point-to-point control, the stage was positioned at eight vertices of an octagon by input shaping, or the current pulse drive. The positioning accuracy by the current pulse drive was better than that by the voltage drive, owing to the large hysteresis of displacement with respect to the applied voltage. The current pulse drive did not excite large vibrations by integrating the current pulses. For continuous path control, the current pulse drive was also compared with the voltage drive by open-loop or proportional integral control. The current pulse drive yielded good performance for a simple open-loop control scenario, and thus can be used for device miniaturization, because in this case no displacement sensors are needed.