Search

High-precision self-localization is one of the basic functions of intelligent vehicles. Moreover, location information is necessary prior information for tasks such as behavioural decision-making and path planning. Recently, the technology of map-based localization has been widely concerned due to its accuracy. A prior map for localization stores one or more environmental features extracted by on-board sensors. These features represent fundamental information about the ground environment such as curbs and lane markings, or the vertical environment such as building outlines. In order to enhance the robustness and positioning accuracy, this paper uses two layers of features to express the environment: one is a bottom layer composed of ground and curb features, and the other is an upper layer composed of a 2D point cloud with vertical features. Firstly, a novel collision-based detection method is proposed to extract the curb features, and a vertical-projection-based detection algorithm is used to detect vertical feature points from multilayer LIDAR. Then, the fusion of a multi-frame feature point cloud is used to represent the rich and complete information on the environment of the intelligent vehicle. Finally, the Monte Carlo localization algorithm is used to obtain an optimal estimate of the vehicle position. The experimental results indicate that the proposed localization algorithm can realize localization accuracy.

High-precision and robust self-localization is one of basic requirements for intelligent vehicles and related applications, because it provides necessary location information for path planning, behavioral decisions. In this paper a localization method based on IBEO LUX scanners, vehicle sensors and priori maps is proposed. The environmental contour features such as trees, street lamps, green belts and building outlines which are fused by the laser scanners and vehicle information are served as localization information. These features are associated with priori feature maps and the optimal vehicle position estimate is obtained by the Monte Carlo Localization framework. Experimental results show that the mean lateral error is less than 10cm and the mean longitudinal error is less than 20cm. So the localization algorithm introduced can meet the requirements of automatic driving demand.

Contemporary importance of the unmanned aerial vehicle (UAV) both for military and civilian applications has prompted vigorous research related with guidance, navigation and control of these vehicles. The potential civilian uses for small low-cost UAVs are various like reconnaissance, surveillance, rescue and search, remote sensing, traffic monitoring, destruction appraisal of natural disasters, etc. One of the most crucial parts of UAVs missions is accurate navigation of the vehicle, i.e. the real time determination of its position, velocity and attitude. Generally highly accurate Strap down Inertial Navigation Systems (SINS) are too heavy to be flown on UAVs. Moreover highly accurate SINS are also highly expensive. Therefore the low-cost and low weight MEMS based SINS with a compromised precision are the viable option for navigation of UAVs. The errors in position, velocity, and attitude solutions provided by the MEMS based SINS grow unboundedly with the passage of time. To contain these growing errors, integrated navigation is the resolution. Complementary characteristics SINS and external non-inertial navigation aids like Global Positioning System (GPS), Celestial Navigation System (CNS) and Doppler radar make the integrated navigation system an appealing and cost effective solution. The non-inertial sensors providing navigation fixes must have low weight and volume to be suitable for UAV application. In this research work GPS, CNS and Doppler radar are used as external navigation aids for SINS. The navigation solutions of all contributing systems are fused using Federated Kalman Filter (FKF). Three local filters are employed for SINS/GPS, SINS/CNS and SINS/Doppler integration and subsequently information from all three local filters is fused to acquire a global solution. Moreover adaptive and fault tolerant filtering scheme has also been implemented in each local filter to isolate or accommodate any undesirable error or noise. Simulation for the presented architecture has validated the effectiveness of the scheme, by showing a substantial precision improvement in the solutions of position, velocity and attitude.

ILS (Instrument landing system) is one kind of the important airborne navigation systems. The performances of the airborne ILS navigation apparatus are vital to safeguard the safe landing of the aircraft. This paper reviews the basic work process of the ILS, summarizes the test theory from the artificial intelligence and virtual instrument view point, discussed the test system construction, and presents the detail design thought. Particular attention is paid to the failures isolating problems. It is shown that test system for airborne ILS navigation apparatus based on artificial intelligence and virtual instrument improved the accuracy and reliability.

On the basis of vehicle automated driving research and the route to record, auto channel identification is particularly important. Therefore, in the case of considering the car speed to quickly identify the road marking line. In this paper, under the condition of considering several roads, as well as the identification of several cameras to do experiments repeatedly. Finally achieved quickly to identify the white mark lines on theroad.

In radar target high-resolution range profile (HRRP) recognition, the target aspect sensitivity problem is mainly solved by angular domain division and template modeling for each angular sector. Semi-parametric statistic model has both advantages of parametric statistic model and non-parametric statistic model, and is proved to be a valid HRRP template. However, when the number of HRRP samples in an angular sector is large, the efficiency might be low because of the non-parametric correction factor used in semi-parametric statistic model. To solve this problem, a semi-parametric statistic model via support vector regression (SVR) is proposed in this paper. By SVR, the idea is to reduce all the samples to only support vectors to formulate the non-parametric correction factor. Experiment results using 5 aircraft HRRP dataset demonstrate the efficiency of the proposed model.

In this paper, a new design method for neural networks is presented based on evolutionary programming. By using an evolutionary algorithm the structure and weights of static neural networks can be simultaneously acquired. This method is further extended to design recurrent neural networks through introducing ‘delayed links’ into networks. Simulation results are also given to illustrate the efficiency of the proposed method.

For extracting mutation information from gear fault signal and achieving a valid fault diagnosis, a gear fault diagnosis method based on morphological mean wavelet transform was designed. Morphological mean wavelet transform is a linear wavelet in the framework of morphological wavelet. Decomposing gear fault signal by this morphological mean wavelet transform could produce signal synthesis operators and detailed synthesis operators. For signal synthesis operators, it was just close to orginal signal, and for detailed synthesis operators, it contained fault impact signal or interference signal and could be catched. The simulation experiment result indicates that, compared with Fourier transform, the morphological mean wavelet transform method can do time-frequency analysis for original signal, effectively catch impact signal appears position; and compared with traditional linear wavelet transform, it has simple structure, easy realization, signal local extremum sensitivity and high denoising ability, so it is more adapted to gear fault real-time detection.

For the research of underwater ship fault detection method in conditions of sailing on the ocean especially in poor visibility muddy sea, a fault detection method under the assist of sonar image processing was proposed. Firstly, did sonar image denoising using the algorithm of pulse coupled neural network (PCNN); secondly, edge feature extraction for the image after denoising was carried out by morphological wavelet transform; Finally, interested regions Using relevant tracking method were taken, namely fault area mapping. The simulation results presented here proved the feasibility and effectiveness of the sonar image processing in underwater fault detection system.

Motion estimation (ME) consumes most of the computational cost in a video encoder system. H.264 adopts seven ME modes and quarter-pel accuracy fractional pixel search, which brings about large scale of computational cost to the fractional pixel ME. To apply H.264 in real-time environment requires not only fast integer pixel ME algorithm but also fast fractional pixel ME algorithms. A novel mode correlation-based fractional pixel motion estimation is proposed in this paper. It makes full use of the correlation of seven ME modes to reduce fractional pel search points in flat homogeneous regions of an image. Experimental results show that compared to the full fractional pixel search, the proposed method can reduce about 50% of fractional pixel search points with negligible degradation in image quality and compression rate. Moreover, this algorithm can integrate other fast integer and fractional pixel ME algorithm to achieve better performance.

We present a photoacoustic imaging system with a linear transducer array scanning in limited-view fields and develop a combined reconstruction algorithm, which is a combination of the limited-field filtered back projection (LFBP) algorithm and the simultaneous iterative reconstruction technique (SIRT) algorithm, to reconstruct the optical absorption distribution. In this algorithm, the LFBP algorithm is exploited to reconstruct the original photoacoustic image, and then the SIRT algorithm is used to improve the quality of the final reconstructed photoacoustic image. Numerical simulations with calculated incomplete data validate the reliability of this algorithm and the reconstructed experimental results further demonstrate that the combined reconstruction algorithm effectively reduces the artifacts and blurs and yields better quality of reconstruction image than that with the LFBP algorithm.