Research On Multimodal Haptic Perception Of Robot Fingertips

The tactile sensor provides information about the shape,elasticity,material properties,and finer texture information of the object,which helps the robot to accurately recognize and manipulate external objects.At present,the demand for tactile sensors in the field of robot fine manipulation is also increasing day by day.However,how to design an efficient and convenient tactile sensing device to obtain the texture and temperature information of objects in real time is still a very challenging problem.In this paper,the method of using the tactile sensing to realize the object attribute perception is studied for the uncertainty in the manipulation of the robot.The tactile sensing device is developed from the two perspectives of texture perception and temperature perception,the Fusion achieves multi-modal perception of the robot.Firstly,a new high-resolution robotic multi-modal tactile sensing device was designed.This device can assist the robot to acquire multi-modal information such as temperature,texture and force of the object,which can be applicable to the robotic two-finger and five-finger hand.The experimental results show that the preparation of the adhesion layer is the key factor affecting the performance of the sensor.It lays the technical foundation for the subsequent texture,temperature sensing and fine operation of the robot.Secondly,an end-to-end texture recognition network algorithm is proposed for objects with different groove textures,so that the robot can perceive the micro texture of the object.A fabric texture data set was established and a texture recognition network based on end-to-end learning was built.Experiments show that for the textile texture database,the proposed end-to-end texture recognition network algorithm has achieved good texture recognition results.On the other hand,a thermochromic material is applied to the tactile sensor,and a tactile sensing device based on temperature recognition is designed by screen printing technology.Experiments show that the sensor has the advantages of fast sensing response speed and large color difference in different temperature intervals.Finally,the texture sensing and temperature sensing sensor performance were integrated into a multi-modal tactile sensing device,and the device was mounted on a Robotiq hand for fine assembly and multi-modal sensing manipulation experiments.Fine assembly experiments show that the tactile sensing device can sense the pose performance of the object in-hand to meet the requirements of the robot's fine assembly experiment.The multi-modal sensing manipulation experiment uses tactile sensing devices with temperature sensing,texture sensing and multi-modal sensing performance to compare the robotic manipulation performance.Experiments show that the manipulation performance of robots using multi-modal tactile sensing devices is much better than that of tactile sensing devices with only a single mode.