Shenshu Xiong
Tsinghua University
Network
Latest external collaboration on country level. Dive into details by clicking on the dots.
Publication
Featured researches published by Shenshu Xiong.
IEEE Transactions on Instrumentation and Measurement | 2005
Guiqiu Bao; Shenshu Xiong; Zhaoying Zhou
Recently, more and more research has been done on micro air vehicles (MAVs). An autonomous flight control system is necessary for developing practical MAVs to be used for a wide array of missions. Due to the limitations of size, weight, and power, MAVs have the very low payload capacity and moments of inertia. The current technologies with rate and acceleration sensors applied on larger aircrafts are impractical to MAVs, and they are difficult to be scaled down to satisfy the demands of MAVs. Since surveillance has been considered as the primary mission of MAVs, it is essential for MAVs to be equipped with on-board imaging sensors such as cameras, which have rich information content. So vision-based techniques, without increasing the MAVs payload, may be a feasible idea for flight autonomy of MAVs. In this paper, a new robust horizon extraction algorithm based on the orientation projection method is proposed, which is the foundation of a vision-based flight control system. The horizon extraction algorithm is effective for both color images and gray images. The horizon can be extracted not only from fine images captured in fair conditions but also from blurred images captured in cloudy, even foggy days. In order to raise the computational speed to meet real-time requirements, the algorithmic optimization is also discussed in the paper, which is timesaving by narrowing the seeking scope of orientations and adopting the table look-up method. According to the orientation and position of the horizon in the image, two important angular attitude parameters for stability and control, the roll angle and the pitch angle, could be calculated. Several experimental results demonstrate the feasibility and robustness of the algorithm.
IEEE Transactions on Instrumentation and Measurement | 2003
Shenshu Xiong; Zhaoying Zhou
In this paper, adaptive filtering approaches of colored noise based on the Kalman filter structure using neural networks are proposed, which need not extend the dimensions of the filter. The colored measurement noise is first modeled from a Gaussian white noise through a shaping filter. The Kalman filtering model of colored noise is then built by adopting an equivalent observation equation, which can avoid the dimension extension and complicated computations. An observation correlation-based algorithm is suggested to estimate the variance of the measurement noise by use of a single layer neural network. The Kalman gain can be obtained when a perfect knowledge of the plant model and noise variances is given. However, in some cases, the difficulties of the correlative method and the Kalman filter equations are the amount of computations and memory requirements. A neural estimator based on the Kalman filter structure is also analyzed as an alternative in this paper. The Kalman gain is replaced by a feedforward neural network whose weight adjustment permits minimization of the estimation error. The estimator has the capability of estimating the states of the plant in a stochastic environment without knowledge of noise statistics. If the noise of the plant is white and Gaussian and its statistics are well known, the neural estimator and the Kalman filter produce equally good results. The neural filtering approaches of colored noise based on the Kalman filter structure are applied to restore the cephalometric images of stomatology. Several experimental results demonstrate the feasibility and good performances of the approaches.
instrumentation and measurement technology conference | 2003
Guiqiu Bao; Zhaoying Zhou; Shenshu Xiong; Xirong Lin; Xiongying Ye
Recently, more and more research has been done on the Micro Air Vehicles (MAVs). Due to the limitation of the size, weight and energy of MAVs, the technologies with sensors on larger aircrafts are not currently available for MAVs. Video-based method is a feasible technique to extract the flight parameter. Pitch angle and roll angle of the MAVs can be extracted based on the horizon in the video images. In this paper, a new robust method to extracting horizon is put forward. The algorithm makes full use of the orientation information of the images. The position and orientation of horizon are obtained in the end. The method cannot only extract the horizon from the fine video images, but also applicable to those unclear images. Compared with the other horizon detection method, the algorithm has less amount of calculations and more robust to the practical situations. The experiment testified that this algorithm is feasible and effective. Our horizon extraction algorithm can correctly identify the horizon in over 99.9% of cases.
international conference on robotics and automation | 2004
Huaiyu Wu; Dong Sun; Zhaoying Zhou; Shenshu Xiong
This paper aims to investigate theoretically and experimentally the dynamic behaviors of the pitch and roll motions of a small-scale unmanned air vehicle in loitering flight. Two fourth-order ARX (AutoRegressive with eXogenous input) models are successfully identified, and the performance analysis is carried out based on the flight test data. The validity of the identified model is verified by both time domain model prediction and frequency domain spectral analysis. With the proposed ARX models, two compensators are further designed using a frequency technique to improve the transient performance of the pitch and roll control channels. Simulations and experiments demonstrate that the proposed ARX model-based compensation control design strategy can improve the flight performance.
international conference on robotics and automation | 2003
Huaiyu Wu; Dong Sun; Zhaoying Zhou; Shenshu Xiong; Xiaohao Wang
This paper presents the development of an electrically powered micro air vehicle (MAV) with a wingspan of 360 mm. A miniature flight control system including a self-made micro video image system especially suitable for MAV is developed. The aerodynamic performance of several airfoil sections at low chords Reynolds number is analyzed in order to find an optimum airfoil section for the MAV prototype. A small-sized propulsion testing setup is built to measure the performance of the motor-gear-propeller-battery combination so that an efficient propulsion system can be obtained. The TH360 MAV with a payload of a self-made micro color video image system has been successfully tested in the real-time flight, where the real-time images of the ground target can be transmitted from the onboard video camera to the ground.
instrumentation and measurement technology conference | 1998
Long Jin; Zhaoying Zhou; Shenshu Xiong; Yun Chen; Minqiang Liu
From the commercial and the technical development standpoints, engineering drawings in electrical computer-aided-design format are more advantageous than those in the traditional paper-based format. With the large stocks of paper drawings in factories and institutes, the demand for conversion is urgent and strong. Currently, several commercial systems are available to convert paper drawings to CAD. However, due to the high complexity and some unpredictable deformation during the processing of drawings, there still exist some problems. Several key solutions to these problems, which are employed in a practical automatic CAD conversion system AVSED, are discussed in this paper. These include the extraction of characters from drawings using a novel text/graphics separation algorithm, complex nonlinear exponential AR (CNEAR) model based character recognition, rapid thinning algorithm and its implementation of hardware, adaptive node regulation algorithm and cross-node tracing algorithm to obtain accurate vectorization of cross lines and neural network based arrow recognition. With these techniques high speed and accurate rate of processing can be achieved in AVSED. The general architecture and algorithms of AVSED are also described. Finally, the AVSED processing results of an original raster drawing is given, and a conclusion is drawn based on the comparison of results by AVSED and a commercial system VPmaxNT.
IEEE Transactions on Instrumentation and Measurement | 2000
Shenshu Xiong; Zhao-Ying Zhou
A complex nonlinear exponential autoregressive (CNEAR) process which models the boundary coordinate sequence for invariant feature extraction to recognize arbitrary shapes on a plane is presented. A neural network structure is constructed to calculate all the CNEAR coefficients synchronically. The network is simple in structure and easy to implement. The nonlinear parameter is easy to determine using the network. The coefficients are adopted to constitute the feature set. They are proven to be invariant to the transformation of a boundary such as translation, rotation, scale, and choice of the starting point in tracing the boundary. Afterwards, the feature set is used as the input to a complex multilayer perceptron (C-MLP) network for learning and classification. Experimental results show that complicated shapes can be recognized with high accuracy, even in the low-order models. It is also seen that the CNEAR model performs better than the complex autoregressive (CAR) model when shapes have random noise on the boundaries or have differentiating features at detailed levels. Finally, an extended training scheme is developed in which the network is gradually retrained sequentially with shapes containing small increments of noise to improve the robustness of the C-MLP classifier.
instrumentation and measurement technology conference | 2004
Zichen Zhu; Shenshu Xiong; Zhaoying Zhou
This paper describes the development of the altimeter for micro aerial vehicles (MAVs) and its application in altitude-holding flight. The potential missions for MAVs include military missions and commercial applications, such as visual reconnaissance, communications relay, search and rescue and field research. For these missions, the altitude control is needed. A micro barometric altimeter was developed as part of the control system to feature altitude hold. The 3-gram altimeter has a high resolution and accuracy. The low frequency noise caused by vibration of airframe is observed, so the Kalman filter is implemented. When this altimeter and PD control law were tested on several unmanned aerial vehicles such as one with 680 mm span, the vehicle succeeded in hovering automatically.
instrumentation and measurement technology conference | 2000
Huaiyu Wu; Zhaoying Zhou; Shenshu Xiong; Wendong Zhang
Functional neuromuscular stimulation (FNS) is one of new approaches in modern rehabilitation engineering. In this article, the adaptive iterative learning control algorithm is investigated for FNS multi-join limb motion system. At first, the FNS control system and its basic principles are given. Then, according to the nonlinear phenomena of the limb muscles, the discrete time-varying nonlinear model with undetermined and unexpected perturbations is defined, and a general expression based on the adaptive iteration learning control algorithm is developed, and the control algorithm structure block diagram is presented. Finally, based on multi-purpose FNS limb motion control system, the clinical experiments on motion trajectory-following of elbow flexion and wrist flexion were conducted by means of the adaptive iteration learning control algorithm and conventional control algorithm. The results of clinical studies have demonstrated that the adaptive iteration learning in control algorithm is more suitable for the improvement of the dynamic response characteristics and the stabilization of limb motion than conventional control algorithm. Furthermore, the stimulated patients have not any bad physiological reactions because the output electrical stimulation pulses generated by the adaptive iteration learning control algorithm vary gently.
instrumentation and measurement technology conference | 2001
Shenshu Xiong; Zhaoying Zhou
An indirect measurement method is proposed for the dynamic parameter estimation of velocity sensors. This approach makes use of data processing abilities of computers and does not require high precision devices. Under the indirect measurement mode, the relationship between the input voltage and the output current of the testing system can be described by a third-order transfer function including physical parameters of the sensor. An identification method for transfer functions with a two step algorithm is presented based on the spectral analysis. The spectral density functions are first obtained from input-output signals of a system, then the dynamic parameters can be estimated from frequency response data. Some parameter estimation algorithms are derived by minimizing a unified loss function with different weight selections and the estimation errors are analyzed. By use of the indirect measurement approach and the modeling algorithms, we can get enough precision results. When the transfer function under the indirect measurement mode is determined, the transfer function in the working state can be obtained through the relationship between the models in the measuring state and in the working state. The methods suggested in the paper can be widely used in the modeling of mechanic-electric systems.