Data processing: measuring – calibrating – or testing – Measurement system – Speed
Reexamination Certificate
2003-06-04
2004-11-09
Bui, Bryan (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system
Speed
C702S141000, C702S142000, C702S145000, C702S150000, C701S045000, C701S072000
Reexamination Certificate
active
06816804
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates generally to a method of sensor fusion that uses a reliability index to economically and accurately estimate a vehicle's absolute velocity.
Absolute vehicle velocity includes both the lateral and longitudinal velocities and is used in many modern vehicle control systems. As manufacturers seek to increase the safety of vehicles, many vehicle control systems need accurate longitudinal and lateral velocity information. These systems include anti-lock brake systems, traction control systems, vehicle stability control systems, by wire control, and interactive or coordinated vehicle motion systems. As vehicle control systems become more complex and exert increasing control over a vehicle's motion, reliable velocity information is needed. Longitudinal velocity is generally determined by measuring the wheel angular speed and calculating the velocity of the vehicle. Wheel speed sensors are common in vehicles with anti-lock brakes or traction control systems.
One problem with using a longitudinal velocity calculated from angular wheel speed is that the calculated longitudinal velocity is not always equal to the actual or true vehicle velocity. For example, in less than optimal conditions, such as when the vehicle is traveling on wet, gravel, snow covered, or icy roads, the potential for the wheels to slip may be increased and therefore, the longitudinal velocity calculated from the wheel speed may differ from the actual longitudinal velocity. To avoid some of these problems, some systems determine the longitudinal velocity of a vehicle from the angular speeds of undriven wheels, but these systems arc also problematic when the vehicle is braking, especially on less than ideal surfaces. Because each vehicle is driven on all-wheel or four-wheel drive vehicles, it is difficult to get an accurate and reliable estimation of the longitudinal velocity, especially during acceleration. To date, the industry has not developed an economically feasible method or system that measures or determines vehicle longitudinal velocity with desired speed, accuracy, and reliability.
Lateral velocity is more difficult to determine than longitudinal velocity because no baseline measurements exist in the lateral direction. Two approaches are generally used to determine lateral vehicle velocity. The first approach uses a vehicle kinetic or dynamic model which requires information about a vehicle's parameters, such as vehicle mass, center of gravity, wheel radius, and wheel cornering stiffness. The second approach uses a kinematic model without considering vehicle dynamics or vehicle parameters.
Use of the vehicle kinetic model generally requires certain assumptions. One such assumption is that the tires of the vehicle are operating within a linear region of known vehicle parameters. For example, kinetic models generally assume a standard tire cornering stiffness. However, this assumption is accurate only for the assumed tire and road surface conditions thereby oversimplifying vehicle dynamics. For an accurate velocity estimation, the method should take into account vehicle dynamics for driving conditions different from the assumed tire and road conditions.
To correct some of the problems associated with lateral velocity estimation based on vehicle dynamics, some methods use a combined state and parameter estimation which attempts to estimate road conditions as well as the lateral velocity. One problem with a combined state and parameter estimation method is that it requires persistent excitation of the desired sensors for parameter convergence. Persistence of excitation is a well known condition required for parameter convergence. More specifically, persistence of excitation is the relation between the richness of the reference signals and the parameter convergence, therefore, the estimated parameters will not converge to the ideal values unless the reference signal or signals satisfy certain conditions. Persistent excitation of the sensors to estimate the vehicle lateral velocity is difficult to achieve because the sensors may not always provide a reading during normal vehicle operation. For example, while at a constant longitudinal velocity with no movement in the lateral direction, the signal received from the accelerometer is generally zero or no signal. Yet another problem is that any signal present during normal vehicle operation may be small and overshadowed by noise inherent in the sensor. Undesirable signal to noise ratios may occur from bumpy, banked, uneven, or slippery road surfaces as well as from vibration inherent in the vehicle and specifically the drivetrain. Another problem with traditional combined state and parameter methods is the need for the last known lateral velocity to estimate the current lateral velocity. If an error is present in the last known lateral velocity, the error generally increases with each cycle.
The second conventional method of determining lateral velocity, the kinematic model, does not consider vehicle kinetic dynamics and vehicle parameters. Kinematic models avoid unknown or time varying vehicle parameters in the estimation algorithm by using a simple kinematic relationship between sensors. The velocity is estimated by integrating accelerometer measurements. However, accelerometer measurements contain many different types of noises. In addition to the exemplary noises given above, the voltage levels or low-frequency component of a signal may drift over time causing the associated signal to be inaccurate. Further, in a kinematic model, the noise is integrated and therefore accumulates estimation errors.
SUMMARY OF THE INVENTION
In view of the above, the present invention is directed to a method and system for vehicle velocity estimation that calculates a noise covariance and uses the calculated noise covariance to determine the reliability of the estimated vehicle velocity.
The present invention effectively combines a vehicle kinetic model and a kinematic model to overcome the above discussed problems while preserving the advantages of each model. To minimize implementation cost, the present invention generally uses available sensors that may already be included in a vehicle or economically be added. Exemplary sensors include a wheel speed sensor, an accelerometer, a yaw rate sensor, and a steering angle sensor.
The system uses signals from the sensors to determine an approximated vehicle velocity, through known techniques. A noise covariance associated with the sensor signals and approximated velocities is then calculated. Selected sensor signals, calculated approximated velocities, and noise covariances are then input into an adaptive Kalman filter framework so that each sensor can be appropriately and systematically combined to result in an estimated vehicle velocity.
In a first embodiment the system for dynamically estimating longitudinal and lateral velocities for use in a vehicle stability control system of a vehicle includes a vehicle sensor producing a signal indicative of a vehicle travel state and a processor communicating with the vehicle sensor. The processor is generally configured to calculate an approximate velocity of the vehicle, determine a noise covariance associated with the sensor signal and approximated velocity, and estimate the velocity of the vehicle using the received vehicle sensor signal, approximated velocity, and the associated noise covariances.
In another embodiment, the system for estimating a velocity of a vehicle for use in a vehicle stability control system is used in a vehicle wherein the system includes an accelerometer for producing an acceleration signal, a wheel speed sensor producing a wheel speed signal, a yaw rate sensor producing a yaw rate signal, a steering wheel angle sensor producing a steering wheel angle signal, a brake position sensor producing a brake position signal, a position sensor producing a throttle position signal, and a processor. The processor communicates with a velocity calculation model which receives said wheel speed sensor signal t
Bui Bryan
Dickinson Wright PLLC
Visteon Global Technologies Inc.
LandOfFree
System and method for estimating velocity using reliability... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for estimating velocity using reliability..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for estimating velocity using reliability... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3319072