Research On Flexible Tactile Sensor Of Intelligent Sensing Machine Hand

Human skin possesses a remarkable sense of touch,enabling precise manipulation of objects and tools.As artificial intelligence and robotics technology advance,it is becoming increasingly important to develop sensor systems that can provide feedback to artificial intelligence and robotics.One of the main challenges for researchers is to create flexible tactile sensors that can mimic human skin,allowing robots to perceive environmental information and safely interact with humans.Tactile sensors should be able to convert physical stimuli generated by contact into measurable electrical signals.Currently,the focus is on developing flexible tactile sensors that can detect various types of physical stimuli,such as compressive stress,bending,stretching,and torque.However,most existing sensors can only respond to a single type of physical stimulus.Intelligent robots often need to process complex and multidimensional tactile information in unstructured environments.Therefore,it is crucial to develop flexible tactile sensors that can sense multiple types of physical stimuli.In summary,the development of flexible tactile sensors capable of sensing multidimensional physical stimuli is essential for advancing intelligent robot technology.These sensors would enable robots to have a more human-like sense of touch and enhance their ability to interact with the world around them.Therefore,the following research works are carried out in this paper:1.Due to the fact that the contact force is a three-dimensional vector,including both magnitude and direction,most pressure sensors currently available can only detect the magnitude of the force without being able to determine its direction.In order to fill this gap,a capacitive three-dimensional force-sensitive flexible tactile sensor capable of simultaneously detecting normal force and tangential force is developed.The sensor design incorporates a three-capacitance sensitive unit structure and utilizes a micro-cone structured elastomer dielectric layer to enhance sensor response performance.A force decoupling method is proposed based on the sensor’s structural design and the mechanical deformation characteristics of the microconical elastomer.The sensitivity of the sensor to normal force reaches 3.5 k Pa^-1 in the low-pressure range of 0-50 Pa,and the sensitivity to tangential force in the range of 0-0.5 N is 0.134 N^-1.These sensitivities are significantly higher than those of sensors using compact dielectric layers without microstructures.Moreover,the microstructured sensors exhibit consistent responses to tangential forces in different directions within the plane.The sensor’s response time is 26 ms,which is faster than the 70 ms neurofeedback delay time of the human tactile system.Additionally,the successful development of a three-capacitance sensor array demonstrates the potential for large-scale array expansion,enabling the detection of pressure distribution over a wide range.2.Based on the structure and signal characteristics of the 3D force flexible tactile sensor,a sensor signal reading circuit is designed using the STM32 microprocessor and the AD7746 capacitor digital conversion chip.This circuit allows for the simultaneous collection of signal changes from the three capacitors during the sensor’s operation.The micro-cone-structured 3D force sensors are assembled on the fingertips of a robot arm’s alloy claw,and the sensor’s signal output is detected in real-time while the robot arm grasps objects.By calculating the capacitance signal using the force decoupling equation,the grip force exerted by the fingertips on the object and the weight of the target object can be accurately determined.This result demonstrates the potential application value of the 3D force sensor in the field of robot haptic feedback.3.When a manipulator performs interactive tasks,the haptic feedback system needs to be able to perceive both the size of the contact force and the change of its own posture.To address this requirement,a dual-mode flexible tactile sensor capable of detecting both pressure and bending signals is developed to simultaneously detect limb bending and contact force during robot movements.The dual-mode sensor consists of a pressure sensing module that is insensitive to bending and a bending sensing module that is insensitive to pressure.The maximum pressure sensitivity of the dual-mode sensor is approximately 0.094 k Pa^-1,and the bending sensitivity is approximately 0.138 rad^-1.The two sensing modules exhibit low crosstalk in dynamic tests.Additionally,the dual-mode sensor demonstrates excellent robustness during repeated pressure and bending tests.4.A“smart glove”with hand motion detection functionality is developed based on bimodal sensors,along with a signal acquisition system developed using an Arduino development board.To simulate the application of the bimodal sensor in a robotic tactile system,the researcher put on the glove and performed different actions,collecting the sensors’output signals under each action.A total of eight actions are designed for the experiment,including four non-contact gesture actions and four object grasping actions,each assigned a specific number.The output signals for different actions are used as the feature values,while the action numbers are used as the predicted values.The decision tree algorithm is elected for action recognition training.The results show that the recognition accuracy using bimodal sensor signals for training is higher than using single pressure or bending module signals.This proves that the bimodal sensor design strategy has broad application prospects in the field of action recognition.