Search Results

Visual Place Recognition (VPR) excels at providing a good location prior for autonomous vehicles to initialize the map-based visual SLAM system, especially when the environment changes after a long term. Condition change and viewpoint change, which influences features extracted from images, are two of the major challenges in recognizing a visited place. Existing VPR methods focus on developing the robustness of global feature to address them but ignore the benefits that local feature can auxiliarily offer. Therefore, we introduce a novel hierarchical place recognition method with both global and local features deriving from homologous VLAD to improve the VPR performance. Our model is weak supervised by GPS label and we design a fine-tuning strategy with a coupled triplet loss to make the model more suitable for extracting local features.

In the intelligent airbag system, the detection accuracy of occupant position is the precondition and plays a vital role to control airbag detonation time and inflated strength during the crash. Through accurately analyzing the seat surface pressure distributions of different occupant sitting position and types, an occupant position recognition approach which purely uses occupant pressure distribution information measured by seat pressure sensors is presented with the method of Support Vector Machine. In the end, the distribution samples with different occupant sitting position and types are used to train and test the recognition approach, and the good validity and accuracy are shown in the experiments.

Accurate sensing of road conditions is one of the necessary technologies for safe driving of intelligent vehicles. Compared with the structured road, the unstructured road has complex road conditions, and the response characteristics of vehicles under different road conditions are also different. Therefore, accurately identifying the road categories in front of the vehicle in advance can effectively help the intelligent vehicle timely adjust relevant control strategies for different road conditions and improve the driving comfort and safety of the vehicle. However, traditional road identification methods based on vehicle kinematics or dynamics are difficult to accurately identify the road conditions ahead of the vehicle in advance. Therefore, this paper proposes an unstructured road region detection and road classification algorithm based on machine vision to obtain the road conditions ahead.

With the development of economic globalization, the speed of development of container terminals is also very rapid. Under the pressure brought by the surge in throughput, the unmanned and intelligent terminals will become the future development direction of terminals. As the cornerstone of the unmanned terminal, the positioning technology provides the most basic position information for system scheduling, path planning, real-time correction, and loading and unloading. Therefore, this paper is aimed to design a low-cost, high-precision, and easy-to-maintain unmanned dock positioning system in order to better solve the problem of unmanned dock positioning. The main research content of this paper is to design a positioning algorithm for unmanned terminal Automated Guided Vehicle (AGV) based on single-line lidar, including point cloud data acquisition, background filtering, point cloud clustering, vehicle position extraction, and result optimization.

Rotating flow inside an internal combustion engine cylinder is deliberately engineered for improved fuel-air mixing and combustion. The details of the rotating flow structure vary temporally over an engine cycle as well as cyclically at the same engine phase. Algorithms in the literature to identify these structural details of the rotating flow invariably focus on locating its center and, on occasion, measuring its rotational strength and spatial extent. In this paper, these flow structure parameters are evaluated by means of complex moments, which have been adapted from image (scalar field) recognition applications to two-dimensional flow pattern (vector field) analysis. Several additional detailed characteristics of the rotating flow pattern - the type and extent of its deviation from the ideal circular pattern, its rotational and reflectional symmetry (if exists), and thus its orientation - are also shown to be related to the first few low-order complex moments of the flow pattern.

Based on the similarity theory and conservation equations, some of the important dimensionless numbers in diesel combustion are deduced and discussed. Existence of similarity is theoretically proved in diffusion (or mixing-controlled) combustion and premixed combustion as well as in spray mixture formation processes in different size diesel engines. With the prerequisite of geometric similarity, scaling rules for some parameters including engine speed, injection pressure and injection duration are established to realize the similarity between large-bore and small-bore diesel engines. To verify the similarity theories, the computational fluid dynamics (CFD) simulation are conducted, and three scaling rules, which keep the engine speed, injection pressure and lift-off length constant, respectively, are compared under the conditions of the light load (0.3 MPa IMEP) and high load (1.55 MPa IMEP) operations.

European standards have set stringent PN (particle number) regulation (6×1011 #/km) for gasoline direct injection (GDI) engine, posing a great challenge for the particulate emission control of GDI engines. Dual-injection, which combines direct-injection (DI) with port-fuel-injection (PFI), is an effective approach to reduce particle emissions of GDI engine while maintaining good efficiency and power output. In order to investigate the PN emission characteristics under different dual-injection strategies, a DMS500 fast particle spectrometer was employed to characterize the effects of injection strategies on particulates emissions from a dual-injection gasoline engine. In this study, the injection strategies include injection timing, injection ratio and injection pressure of direct-injection.

This paper summarized the development methodology and technical experiences on Formula Student racecar engines acquired by Jilin University from 2011 to 2015. This series of engines are all based on 600cc 4-cylinder motorcycle gasoline engines and were modified to turbocharged engines which met the Formula Student technical regulations, in order to achieve higher power output, wider torque band as well as lower fuel consumption. During the development process, multiple research projects have been conducted surrounding the turbocharging technology. These research projects have covered multiple areas including the matching of the flow rate characteristics of the engine and the turbocharger, the design of intake and exhaust systems, research on the wastegate as well as its actuator, the tuning and control of the boost pressure as well as the design of the lubrication system for the turbocharger, etc.

In this paper, in order to avoid the frequent switching of engine operating points and improve the fuel economy during driving, this paper proposes a control strategy for the 4-wheel drive (4WD) hybrid vehicle based on wavelet transform. First of all, the system configuration and the original control strategy of the 4WD hybrid vehicle were introduced and analyzed, which summarized the shortcomings of this control strategy. Then, based on the analyze of the original control strategy, the wavelet transform was used to overcome its weaknesses. By taking advantage over the superiority of the wavelet transform method in multi signal disposition, the demand power of vehicle was decomposed into the stable drive power and the instantaneous response power, which were distributed to engine and electric motor respectively. This process was carried out under different driving modes.

This work investigates the suitability of n-butanol for enabling PCCI, HCCI, and RCCI combustion modes to achieve clean and efficient combustion on a high compression ratio (18.2:1) diesel engine. Systematic engine tests are conducted at low and medium engine loads (6∼8 bar IMEP) and at a medium engine speed of 1500 rpm. Test results indicate that n-butanol is more suitable than diesel to enable PCCI and HCCI combustion with the same engine hardware. However, the combustion phasing control for n-butanol is demanding due to the high combustion sensitivity to variations in engine operating conditions where engine safety concerns (e.g. excessive pressure rise rates) potentially arise. While EGR is the primary measure to control the combustion phasing of n-butanol HCCI, the timing control of n-butanol direct injection in PCCI provides an additional leverage to properly phase the n-butanol combustion.

Three-phase sequential turbocharging system with two unequal-size turbochargers is developed to improve fuel economy performance and reduce emission of the automotive diesel engine, which satisfies wide range of intake flow demand. However, it results in complicated transient control strategies under frequently changing operating conditions. The present work aims to optimize the control scheme of boost system and fuel injection and evaluate their contributions to the improvement of transient performance. A mean value model for diesel engine was built up in SIMULINK environment and verified by experiment for transient study. Then a mathematical model of optimization issue was established. Strategies of control valves and fuel injection for typical acceleration and loading processes are obtained by coupled calculating of the simulation model and optimization algorithm.

Flash boiling spray has been proven to be a useful method in providing finer fuel droplet and stronger evaporation in favor of creating a homogeneous fuel-air mixture. Combustion characteristics of flash boiling spray are thus valuable to be investigated systematically for aiding the development of efficient internal combustion system. An experimental study of flash boiling spray combustion in a SIDI optical engine under early injection has been conducted. The fuel, Iso-octane, was used across all tests. Three fuel spray conditions experimented in the study: normal liquid, transitional flash boiling and flare flash boiling sprays, within each case that Pa/Ps ratio was set in (>1), (0.3~1), and (<0.3) respectively. A small quartz insert on the piston enables optical access for observing combustion process; non-intrusive measurements on flame radicals has been carried out using a high-speed color camera.

Lift-off length is defined as the distance from injector hole to the location where flame stabilized on a high injection pressure direct injection (DI) diesel spray. In this paper we used the high-speed (40 kHz) Schlieren and time-averaged OH chemiluminescence imaging technique to simultaneously measure the flame lift-off locations on a DI diesel spray in an optically accessible and constant-volume combustion vessel. The time-resolved development of the diesel spray acquired from the high-speed Schlieren imaging system enabled us to observe the instantaneous spray structure details of the spray flames. The OH chemiluminescence image obtained from a gated, intensified CCD video camera with different delay and width settings was used to determine the quiescent lift-off length. Experiments were conducted under various ambient temperatures, ambient gas densities, injection pressures and oxygen concentrations.

In the face of energy and environmental problems, how to improve the economy of fuel cell vehicles (FCV) effectively and develop intelligent algorithms with higher hydrogen-saving potential are the focus and difficulties of current research. Based on the Toyota Mirai FCV, this paper focuses on the short-term speed prediction algorithm based on the back propagation neural network (BP-NN) and carries out the research on the short-term speed prediction algorithm based on BP-NN. The definition of NN and the basic structure of the neural model are introduced briefly, and the training process of BP-NN is expounded in detail through formula derivation. On this basis, the speed prediction model based on BP-NN is proposed. After that, the parameters of the vehicle speed prediction model, the characteristic parameters of the working condition, and the input and output neurons are selected to determine the topology of the vehicle speed prediction model.

In order to meet increasingly stringent emission regulations, it is significance to establish a control- oriented transient NOx and PM emission prediction model and improve the accuracy and real-time performance. In this study, the prediction model of transient PM emissions based on Transformer is established. In terms of model accuracy and real-time performance, Transformer emission prediction model is compared with Multilayer perceptron (MLP) neural network and Long-Short Term Memory (LSTM) emission prediction model. The results show that the performance of Transformer transient emission prediction model is superior to other model structures, it can be used for real-time prediction.

To reach the goal of optimal performance match between engine and transmission, the dynamic characteristics of engine should be taken into consideration. In the paper, the dynamic torque and fuel consumption models of engine, described by a multi-layers feed forward neural network, were established. Based on that, the methods used to calculate the optimal dynamic and economical shift schedules with dynamic 3-parameters were put forward. The shift schedule with dynamic 3-parameters based on neural network model is proven to be superior to the shift schedule with only 2-parameters in both dynamic performance and fuel economy by the test.

With the rapid development of modern economy and society, traffic congestion has become an increasingly serious problem. Vehicle cooperative driving can alleviate traffic congestion and improve road traffic capacity. Compare with vehicle separate control, cooperative driving combines various vehicle systems, and highly integrates information on obstacle location, vehicle status and driving intention. Then the controller uniformly issues instructions to ensure the orderly driving of the platoon. In the cooperative driving platoon, the displacement difference and the speed difference between vehicles have a certain relationship, which reduces the possibility of traffic accidents and then improves the safety of driving. In the process of cooperative driving, if there are multiple vehicles whose speeds don’t meet the current lane requirements, or if there are obstacles ahead, multi-vehicle lane change measures must be taken.

With the aim of addressing the problem of inconsistency of the traditional proportion integration (PI) driver model with the actual driving behavior, a longitudinal driver model based on the elastic net regression (ENR) and adaptive network fuzzy inference system (ANFIS) method is proposed. First, longitudinal driving behavior data are collected through bench tests to extract the characteristic parameters that affect driving behavior. A quadratic regression model is established after considering the nonlinear characteristics of the driver behavior. The multi-collinear problem of high-dimensional variables in the regression model is solved by the ENR method, and the parameters with significant influence on driving behavior selected. A longitudinal driver model of ANFIS was established with the selected characteristic parameters as input. Finally, the validity of the model is verified by comparing it with the PI and ENR driver models.

In the field of automatic driving, the combination of 3D LIDAR and inertial measurement unit (IMU) is a common sensor configuration scheme in laser point-cloud localization, high-precision map making and point-cloud target detection. So it is critical to calibrate LIDAR and IMU accurately. At present, due to the large volume and high cost of 3D LIDAR with high-line-number(Such as 64 lines or 128 lines), the configuration scheme of using multiple low-line-number 3D LIDARs appears in the automatic driving vehicle sensing system. However, the common calibration methods are not suitable for multi 3D LIDARs and IMU parameters calibration on autonomous vehicle, which have the disadvantages of cumbersome implementation and low accuracy. In this paper, a joint calibration test platform composed of dual LIDARs and IMU is assembled, and a method of precise automatic calibration based on GPS/RTK data is proposed.

A novel method is applied to analysis the autoignition phenomenon. Experiments on the study of autoignition characteristics of diesel fuel were carried out with a Controllable Active Thermo-Atmosphere Combustor. The results show that the method for autoignition studying of liquid fuel is of feasibility. Autoignition delay time and autoignition height from the nozzle increase with the coflow temperature decreasing and autoignition delay time changes sensitively under lower coflow temperature. Liftoff height of diesel spray flame decreases with the increasing of coflow temperature. Lower temperature causes higher variance of liftoff height. It might be speculated that there are two different mechanisms of flame stabilization that the lower lift-off heights flames are related to a balance between the flow velocity and flame speed while the higher lift-off heights flames are stabilized by the mixture autoignition.