Particle sampling method and sensor fusion and filtering method

• US 7,379,844 B2
• Filed: 02/15/2006
• Issued: 05/27/2008
• Est. Priority Date: 02/17/2005
• Status: Expired due to Fees

• Current Assignee: Sungkyunkwan University Foundation For Corporate Collaboration (SungKyunKwan University (SKKU))
• Original Assignee: Sungkyunkwan University Foundation For Corporate Collaboration of Sungkyundwan University (SungKyunKwan University (SKKU))
• Inventors: Lee, Jang-Yong, Baek, Seung-Min, Lee, Seong-Soo, Lee, Suk-Han
• Primary Examiner(s): Barlow, Jr.; John E
• Assistant Examiner(s): Le; Toan M
• Application Number:
  US11/276,127
• Publication Number:
  US 20060190217A1
• Time in Patent Office: 832 Days
• Field of Search: 702/181, 702/189, 702/190, 703/2, 700/29, 700/55, 708/300
• US Class Current: 702/181
• CPC Class Codes: G05B 13/04 involving the use of models or simulators
• CPC Class Codes: G05B 17/02 electric

Abstract

Disclosed is a technique for obtaining an estimate and variance of each variable based on a constraint manifold. Particles (or samples) are sampled in order to filter and fuse ambiguous data or information on at least one state variable of a system using the particles. The sampling is carried out in consideration of an influence which non-linearity of the constraint manifold of a system model, an observation model or another system model exerts on a probability distribution of the state variable. With this construction, it is possible to reduce decrease of fusion and filtering performance, decrease a Gaussian approximation error, and detect mismatched information.

Claims

1. A particle sampling method for sampling particles in order to filter ambiguous data or information on at least one state variable of a system using the particles, characterized in that

sampling is carried out in consideration of the influence of the non-linearity of system dynamic model, observation model and/or other system constraints, on the probability distribution of state variables, and the particle sampling method comprises the steps of:

when the sampling performed on a constraint manifold in a hyper space of an arbitrary model equation at regular intervals, where a constraint manifold represents a system dynamic model and/or other system constraints defined in the hyper space of current and previous state variables, wherein the hyper space of an arbitrary model equation at regular intervals is defined as “

Uniform Sampling on Constraint Manifold,”

previously performing the sampling in the hyper space from numerous sample meeting an equation for the system model at regular intervals, and obtaining particles of a previous state variable and a current state variable;

when an interval divided uniformly on an axis of the previous state variable is defined as a bucket, finding a weight of each particle of the current state variable estimated through a weight allocated by the particles of the previous state variable and prior probability information of the previous state variable and through a number of the particles existing in the bucket;

finding a prior probability distribution of a current state estimated from the estimated weight of the sampled particles of the current state variable and from the estimated weight of each particle of the current state variable; and

previously performing the sampling in the geometrical space from the numerous samples meeting an equation for the observation model at regular intervals, and obtaining the particles of the previous state variable and particles of an observation variable.

2. The particle sampling method according to claim 1, characterized by the step of obtaining the samples which meet the non-linearity of the constraint manifold of the system model, the observation model and/or the other system constraints by a Monte Carlo method.

3. The particle sampling method according to claim 1, characterized in that, when the number of the samples existing in each bucket is denoted by m, the weight of the particles of the current state variable is found by:

Weight of particles of the current state variable=Weight of samples existing in each bucket/m.

4. A fusion and filtering method for fusing and filtering ambiguous data or information on at least one state variable of a system using particles, characterized by the steps of:

setting a data or information fusing and filtering space composed of state and observation variables;

defining a constraint manifold having various constraints in the fusing and filtering space;

calculating a joint probability distribution on the constraint manifold using a prior probability distribution of the variables;

calculating a marginal probability distribution of the variables from the joint probability distribution; and

obtaining estimate and variance of each of the variables.

5. A particle sampling method for sampling particles in order to filter ambiguous data or information on at least one state variable of a system using the particles, characterized in that

sampling is carried out in consideration of the influence of the non-linearity of system dynamic model, observation model and/or other system constraints, on the probability distribution of state variables, and the particle sampling method comprises the step of:

obtaining a weighted particle distribution of a current state variable by uniformly sampling the particles on an axis of a previous state variable in such a way as to find, from the previous state variable, the effect of a system dynamic model on the probability distribution of a current state variable with the weights, referred to here as prior probability weights, assigned to the uniformly sampled particles based on the prior probability distribution of the current state variable; and

finding the final weights of the particles of a current state variable, where the particles are from the above uniform sampling with weights, by applying the probability distribution of the observation variable from measurement to the weighted particles of the current state variable in accordance with an equation for the observation model, such that the observed probability distribution and the prior probability weights are combined into the final weights.

6. A particle sampling method for sampling particles in order to filter ambiguous data or information on at least one state variable of a system using the particles, characterized in that

sampling is carried out in consideration of the influence of the non-linearity of system dynamic model, observation model and/or other system constraints, on the probability distribution of state variables, and the particle sampling method comprises the steps of:

obtaining a particle distribution of a current state variable by aggregating, in a proper proportion, a particle distribution from uniformly sampling the particles on an axis of a previous state variable in such a way as to find, from the previous state variable, the effect of a system dynamic model on the probability distribution of a current state variable and a particle distribution of a current state variable from Monte Carlo sampling the particles on an axis of a previous state variable based on a prior probability distribution of the current state variable, wherein the prior probability distribution is estimated from an estimated weight of the sampled particles of the current state variable and from an estimated weight of each particle of the current state variable; and

finding the weights of the particles of the current state variable, where the particles are from the above aggregation of two particle distributions, by applying the probability distribution of the observation variable from measurement to the particles of a current state variable in accordance with an equation for the observation model.

7. The particle sampling method according to claim 6, characterized by the step of obtaining the samples which meet the non-linearity of the constraint manifold of the system model, the observation model and/or the other system constraints by a Monte Carlo method.