Pohang University of Technology in South Korea: Research Status and Algorithm based Performance Enhancement of Robot Tactile Array Sensors

Automated robots have achieved significant industrial success in executing predetermined work plans, while the recent emergence of advanced collaborative and service robots has expanded the potential applications of robots and is expected to usher in the era of humanoid robots.The complex motion control of these robots has been achieved through traditional motion control sensors (such as torque), such as sensors suitable for joint motors, internal measurement units (IMUs) for measuring acceleration and angular velocity, and force/torque integrated sensors for complex tasks (such as changing and maintaining posture).Although collaborative robots have made significant progress in repetitive specific tasks, there are still technical challenges in applying them to general tasks, including direct real-time and precise sensing of external mechanical stimuli (force, mass, center of mass, friction, torque, strain, etc.), recognition of forces applied from various directions, identification of material types and characteristics (surface texture, hardness), perception of multimodal stimuli (temperature, force), etc.In recent years, the research trend of robot tactile sensors has shifted from rigid sensors manufactured in microelectromechanical systems manufacturing processes to flexible/stretchable sensors manufactured on elastic substrates to mimic human skin functions. So far, identification of various force sensing mechanisms has been reported, including piezoresistive, capacitive, inductive, piezoelectric, and frictional. Recently, the addition of an ion sensing layer to electronic signal platforms has been proposed as a new type of electronic tactile sensor.

Recently, HyeYeon Yang from Samsung Advanced Technology Institute and Unyong Jeong from Pohang University of Technology in South Korea summarized the latest developments in robot tactile sensors and their algorithmic applications.

In this review, the team first elaborated on the performance parameters of tactile sensors. The human skin responds to tactile stimuli through four mechanoreceptors: slow adaptation receptors (SA-I and SA-II) that respond to static pressure, and fast adaptation receptors (FA-I and FA-II) that respond to dynamic pressure or vibration.In contrast, robot hand tactile perception does not require imitating the full performance of human hands, nor does it require achieving human level resolution in all areas. Therefore, robot tactile sensors should consider several parameters: sensitivity of signal linearity, large dynamic sensing range, no or minimum signal lag, and high spatial resolution.

Afterwards, the team introduced the reported tactile sensor sensing mechanisms, including resistive, capacitive, inductive, piezoelectric, and frictional sensors. Although significant progress has been made in device tactile sensors, some types of which can rival or even surpass human levels, single device tactile sensors still face limited spatial resolution limitations and are difficult to apply in a wide range of scenarios.Therefore, the team also discussed the research progress of array sensors. Different array configurations and architectures can lead to different resolution improvements, but they can also cause crosstalk, noise, and complexity in the fabrication process.

Finally, the team discussed commonly used signal processing algorithms in tactile sensors, including dimensionality reduction algorithms, time-frequency analysis algorithms, and classification algorithms, and discussed the recognition efficiency, classification/prediction accuracy, robustness, and universality of the algorithms.The author points out that the current methods for practical robot tactile sensors are carried out in two directions: first, based on materials and parts to ensure high-quality sensors, and second, based on algorithms to improve the functionality of simple structured sensors. In addition, with a focus on the next generation of intelligent robots, the integration of multiple technologies, such as tactile, visual, and auditory signals, should also be considered.

Source: Sensor Expert Network