Jianmin Duan

A method of spatial calibration for camera and radar

The technique of multi-sensor fusion has already been used widely in intelligent vehicle environment perception. The spatial calibration of the radar and camera is the basis for road detection with information fusion real-timely. So, a method of spatial calibration of radar and the camera has been proposed to implement the spatial calibration of the two sensors. The non-linear distortion of the camera is considered in the paper. Coordinate transformation between the coordinates relevant to the radar and camera is introduced as the constraints. After transforming the feature between the related coordinates of image and the relative coordinates of radar, the calibration parameters are determined with the least square error function. Experiment results indicate that this method is simple and easy to realize, which has the high accuracy that satisfy requirement of the system.

Machine-vision based preceding vehicle detection algorithm: A review

Onboard vehicle detection system is of great importance to reduce vehicle collision accident and increase the driving safety on road. It aims at detecting vehicles appearing around the ego vehicle using vehicle-mounted camera, so as to alert the driver about driving environments and possible collision with other vehicles. In this paper, we analyze the detail difficulties lying in the problem and review most of the literatures. A typical vehicle detection algorithm includes two steps: hypothesis generation and hypothesis verification. After a vehicle is detected, its tracked. This paper introduces the principle of typical methods of detection and tracking and analyzes their respective pros and cons. Finally, we propose some research directions in the future.

Lateral and longitudinal dynamic coupling control for vehicles lane changing

There are strong dynamic coupling effects exist between vehicles lateral and longitudinal motion, especially in lane changing process. It is difficult to get ideal control effect by lateral and longitudinal control separately. Thus, to realize a smooth and safety lane changing process the nonlinear lateral and longitudinal coupling vehicle model was established. According to sliding mode control and disturbance observer theory the integrated longitudinal and lateral control for lane change was studied and a composite control strategy was proposed based on sliding mode control and disturbance observer. This strategy used disturbance observer to estimate outside disturbance or uncertainty that exist in system and the estimate value was to compensated the sliding mode controllers switching gain for improve controllers robustness and weaken its output chattering. Prove the stability of this control strategy based on Lyapunov theory. The simulation results of an example are investigated to show the feasibility and the effectiveness of this control strategy.

AC electric power steering system modeling with Feed-forward fuzzy control algorithm

In this paper, using the permanent magnet synchronous motor (PMSM) as the motor of electric power steering system; establish mathematical models of the system, the completion of the EPS system architecture design. Adopt the composite method of Feedforward and the fuzzy PI control track on the target current, its improve the speed and accuracy response of system, reducing the tracking delay. Combine with steering system model, two freedom degrees vehicle model designing the power curve of control algorithm and simulating the overall system. The results of simulation show that the design of the curved power algorithm achieves to a good target of power control, and the system is stability.

Machine vision based localization of intelligent vehicle

Aiming at implementing the lane recognition and determining the host vehicles parameters of position and direction accurately according to lane line parameters, realize the lane departure warning, one kind of practical algorithm for lane recognition and the host vehicle localization has been proposed. In the algorithm, lane parameters are obtained using Hough Transformation. Combined the inner and outer parameters of CCD and lane parameters, the parameters such as the host vehicles location parameter, the direction variable and the lane width are obtained using the coordinate transformation. The experiments indicated that this algorithm has the good adaptive ability and anti-jamming ability when roadway structure and illumination condition changes. It can satisfy many kinds of initiative safety systems requests in a certain extent.

Research on Lane Detection Based on Machine Vision

In order to enhance the real-time and stability of lane detection based on machine vision, a method for line detection based on combined road model is proposed. After image classification according to illumination, the image is processed by different algorithms during image pretreatment. For straight line and curve line in the vicinity, improved Hough Transform (HT) is adopted for line detection and tracking. Both of the road boundaries are fitted using Catmull-Rom Splines based on control points search algorithm for curve line in the distance. For various kinds of lanes on most structural road, experiment results indicate that the method has good robustness and stability.

The Constructing of Multi-Agent Intelligent Control System for Behaviors Coevolution

To attain adaptive online learning capability of behaviors coordination and cooperation for the distributed intelligent control system which is situated in variational and open environment, the coevolution technique was introduced. It adopted the method of decomposing the complex distributed control system into basic behavior modules, and constructed them into the multi-agent system whose behaviors can coevolve. The distributed parallel coevolution learning mechanism and its algorithms were designed. It makes each basic behavior rule-base and global behavior coordination rule-base coevolve, consequently, the local control and global behavior coordination obtain adaptive online learning capability. The system is simple and flexible, and may be applied to embedded distributed control system operating at limited resource