In order to provide useful information for grasping of agricultural robot via perceiving surface roughness of fruits and vegetables, we designed and produced a PVDF tactile sensor, perceiving surface roughness via processing of tactile signal. With the mechanical analysis by ANSYS, we could determine the effective area and the reasonable location of sensor model. Four PVDF piezoelectric films and the resistance strain chips were randomly arranged in the model of fingers. The random distribution of the sensor elements would make tactile sensor as much as possible to obtain a wealth of information about the work environment. We built up a platform of tactile information detection to process data collection from three different roughnesses of fruits through multi channels data acquisition program. After extracting the sample characteristics, the maximum value, the minimum value and the difference between the maximum and minimum values of the signal were calculated as the tactile sensor characteristics. The differences between the maximum value and the minimum value of the signal were used as the pressure sensor characteristics. Then the SVR algorithm model was established to test the surface roughness of the fruits and vegetables. Experimental results were consistent with the actual setting of roughness class, which proved that the tactile sensor can detect surface roughness characteristics of fruits and vegetables.