Location determination using a state space estimator

A system comprising:

• a mobile device, and
• a non-transitory storage device storing instructions operable to cause the mobile device to perform operations comprising:
• determining an initial location of the mobile device, the initial location being at a venue, the venue comprising a space accessible by a pedestrian and one or more constraints of movement of the pedestrian;
• determining, based on a reading of a sensor of the mobile device, a state of motion of the mobile device;
• determining a plurality of candidate locations of the mobile device based at least in part on the initial location, the state of motion, and a map of the venue specifying the one or more constraints, wherein determining the candidate locations comprises:
• upon determining that the mobile device is stationary, generating the candidate locations based on a state space system noise model;
• or
• upon determining that the mobile device is moving, generating the candidate locations that avoids moving the mobile device into the constraints;
• receiving a location observation, the location observation including an estimated location of the mobile device in the space, the estimated location being determined based on readings of one or more sensors of the mobile device and location fingerprint data describing attributes of signals measurable by the one or more sensors when the mobile device is at the venue;
• filtering the candidate locations using the location observation, including determining a likely location of the mobile device based on the candidate locations, the observation, and a probability model, wherein:
• the probability model comprises a particle filter;
• and
• filtering the candidate locations comprises feeding the candidate locations and the observation as inputs to the particle filter, wherein the particle filter estimates a probability density of the candidate locations using a previous location of the mobile device conditioned upon the observations and filters the candidate locations according to the probability density;
• and
• providing the likely location for output as a current location of the mobile device.

A non-transitory computer-readable medium storing instructions operable to cause one or more processors to perform operations comprising:

• determining, by a mobile device, an initial location of the mobile device, the initial location being at a venue, the venue comprising a space accessible by a pedestrian and one or more constraints of movement of the pedestrian;
• determining, based on a reading of a sensor of the mobile device, a state of motion of the mobile device;
• determining a plurality of candidate locations of the mobile device based at least in part on the initial location, the state of motion, and a map of the venue specifying the one or more constraints, wherein determining the candidate locations of the mobile device comprises:
• determining a plurality of random locations that are reachable by the mobile device under constraints of the state of motion and the one or more constraints;
• and
• designating the random locations as the candidate locations;
• receiving, by the mobile device, a current location observation, the current location observation including an estimated location of the mobile device in the space, the estimated location being determined based on readings of one or more sensors of the mobile device and location fingerprint data describing attributes of signals measurable by the one or more sensors when the mobile device is at the venue;
• filtering the candidate locations using the initial location and the current location observation, including determining a likely location of the mobile device based on the initial location, the candidate locations, the observation, and a probability model, wherein:
• the probability model comprises a particle filter;
• and
• filtering the candidate locations comprises feeding the initial location, the candidate locations, and the current location observation as inputs to the particle filter, wherein the particle filter estimates a probability density of the candidate locations using the initial location of the mobile device conditioned upon the observations and filters the candidate locations according to the probability density, the initial location being estimated from a prior location observation that was received by the mobile device at a pre-specified duration prior to receiving the current location observation;
• and
• providing the likely location for output as a current location of the mobile device.