×

Location determination techniques

  • US 20070149216A1
  • Filed: 12/05/2006
  • Published: 06/28/2007
  • Est. Priority Date: 12/07/2005
  • Status: Active Grant
First Claim
Patent Images

1. A method for estimating a set of properties of a target object in an environment, wherein the set of properties comprises location;

  • the method comprising;

    modelling the environment with a topology model and a data model, wherein the topology model indicates permissible locations and transitions within the environment, and the data model indicates at least one location-dependent physical quantity for each of several permissible locations indicated by the topology model;

    modelling location changing characteristics of the target object with one or more motion models, wherein each motion model models a specific type of a target object and obeys permissible locations and transitions indicated by the topology model;

    associating to the target object one or more co-located sensing devices, each of which is capable of making observations of one or more of the location-dependent physical quantities;

    assigning to the target object a set of particles, each of which has a set of attributes, wherein the set of attributes comprises at least a location in relation to the topology model;

    estimating the set of properties of the target object with the set of attributes of the particles assigned to the target object; and

    updating the set of particles in a plurality of update cycles, wherein each update cycle comprises the following phases a) to c);

    a) determining a degree of belief for each particle to accurately estimate the set of properties of the target object, using the data model and observations from at least one sensing device associated to the target object;

    b) determining a weight for each particle based on at least the determined degree of belief; and

    c) generating a set of new particles for update cycle n+1 wherein;

    at least some of the new particles are based on one or more parent particles for update cycle n, wherein the likelihood of a particle for update cycle n to be selected as a parent particle for a new particle in update cycle n+1 is a non-decreasing function of the weight of the particle; and

    the set of attributes of a new particle for update cycle n+1 is derived from the set of attributes of one or more parent particles for update cycle n by using at least one of the one or more motion models and a predetermined algorithm.

View all claims
  • 4 Assignments
Timeline View
Assignment View
    ×
    ×