3D shape estimation of vertical distance using ultrasonic sensors for the preview control of an active suspension vehicle

Abstract

Recently, most automobile companies apply an active suspension system to commercial luxury vehicles gradually. Also, they are using a stereo camera to predict the shape of road surface, and doing the preview control of active suspensions based on the estimated shape of road surface. The preview control of active suspensions can provide better ride quality by compensating the slow actuation ability of active suspensions. However, a high-performance processor should be needed if a stereo camera is used for the preview control of active suspensions because, basically, there is so much data by the characteristics of a camera which is a camera is based on surface detection. And, a camera is vulnerable of the variation of weather and light intensity and the cost of a camera system including a camera and a processor is more expensive than that of other short-range detection sensor systems based on an ultrasonic sensor and an infrared sensor. Thus, this paper proposes an estimation technology of a 3D vertical distance between an ultrasonic sensor module and a road which can be applied for the preview control of an active suspension vehicle based on ultrasonic sensors which system have fewer data that should be processed, and which are cheaper than a stereo camera system which is mainly used for the preview control of an active suspension vehicle. A 3D shape of vertical distance between an ultrasonic sensor module and a road is composed of the 2D shape of vertical distance that is the result of the distance measurement using ultrasonic sensors and the additional time axis that describes the variation of a 2D shape of vertical distance. This paper introduces two different methods to estimate a 3D shape of vertical distance between an ultrasonic sensor module and a road based on Time of Flight (TOF) which can be applied for the preview control of an active suspension. The 2D shape of vertical distance which excludes the time axis from a 3D shape of vertical distance between an ultrasonic sensor module and a road is formed by interpolation using measured distances of two points in the first method. Two ultrasonic transmitters uses an ultrasound of only the 40 kHz frequency component. The signal interference (cross talk) happens when measuring the distance of two points using an ultrasound of only one frequency component. Thus, the alternate distance measurement for two points is proposed in the first method. First method uses the threshold method to determine TOF. It adopts the trilateration method which is widely used for Global Positioning System (GPS) to enhance the reliability of distance measurement and correspond to the fail situation of ultrasonic receivers. Also, it utilizes Rule-based filter and Low Pass Filter (LPF) to filter TOF data and improve the accuracy of distance measurement. The first method emphasizes on the 3D shape estimation of vertical distance with a logic as light as possible and reliability of distance measurement in the consideration of the cost side of the estimation system. Two ultrasonic transmitters which use around 35 kHz and 45 kHz frequency component respectively are utilized in the second method. The distance of two points of the two ultrasonic transmitters and one additional point through frequency identification and transmitter classification is measured. Then, a 2D shape of vertical distance which excludes the time axis from a 3D shape of vertical distance between an ultrasonic sensor module and a road is formed by interpolating using measured distances of three points in the second method. The second method adopts the cross-correlation method to separate each frequency component. Also, it uses the Hilbert transform and interpolation methods to enhance the accuracy of distance measurement. The second approach emphasizes on improvement of the accuracy of a 3D shape estimation of vertical distance by adding one more measurement point by frequency identification. The first method is verified by comparing a 3D shape data of vertical distance between an ultrasonic sensor module and a road with ultrasonic sensors and a laser scanner for reference distance measurement attached on a real vehicle. The experiment is conducted on a dry asphalt road and urban condition with a vehicle velocity 40 km/h. And, the second method is verified by comparing a 3D shape data of vertical distance which is interpolated using two points of the two ultrasonic transmitters and one additional point through frequency identification and transmitter classification based on ultrasonic sensors and a 3D shape data of vertical distance using a laser scanner on a test bench which is installed with a model of a bump with sensors moving slowly.