High-performance inertial measurements using a redundant array of inexpensive inertial sensors

Patent metadata

Field	Value
Publication number	US-11378399-B2
Application number	US-201615760033-A
Country	US
Kind code	B2
Filing date	Sep 14, 2016
Priority date	Sep 14, 2015
Publication date	Jul 5, 2022
Grant date	Jul 5, 2022

Abstract

A method for determining the rotational rate of a movable member using an array of inertial sensors is provided. The method includes defining a hidden Markov model (“HMM”). The HMM represents a discrete value measurement of the rotational rate of the movable member. A transition probability of the HMM accounts for a motion model (linear or non-linear) of the movable member. An observation probability of the HMM accounts for noise and bias of at least one of the inertial sensors of the array of inertial sensors. A processor receives input from the array of inertial sensors. The processor determines the rotational rate of the movable member by solving for an output of the HMM using the input received from the array of inertial sensors. The processor may use a forward algorithm, a forward-backward algorithm, or a Viterbi algorithm to solve the HMM.

First claim

What is claimed is:

1. A method for determining rotational rate of a movable member using an array of inertial sensors, the method comprising:
defining a hidden Markov model (“HMM”), wherein hidden states of the HMM represent a discrete value measurement of the rotational rate of the movable member, and transition probability of the HHM accounts for a motion model of the movable member, and observation probability accounts for noise and bias of at least one of the inertial sensors in the array of inertial sensors;
receiving, by a processor, input from the array of inertial sensors;
determining, by the processor, the rotational rate of the movable member by solving for an output of the HMM using the input received from the array of inertial sensors; and
controlling, by the processor, motion of the movable member using the rotational rate of the movable member.

2. The method of claim 1 , wherein the hidden states represent a plurality of angular rates measured at a plurality of time steps.

3. The method of claim 1 , further comprising:
applying a pre-filter to obtain an approximate of the rotational rate of the movable member; and
using the approximate rotational rate to discretize a state space.

4. The method of claim 1 , further comprising:
maintaining as a constant a state space size as the number of inertial sensors of the array of inertial sensors increases.

5. An apparatus comprising:
a platform, wherein the platform comprises
an array of inertial sensors coupled to the platform, and
a processor coupled to the platform that receives inputs from each inertial sensors of the array of inertial sensors; and

a movable member coupled to the platform, wherein
each inertial sensor of the array of sensors outputs a rotational rate of the movable member,
the processor defines a hidden Markov model (“HMM”), wherein hidden states of the HMM represent a discrete value measurement of the rotational rate of the movable member, and transition probability of the HHM accounts for a motion model of the movable member, and observation probability accounts for noise and bias of at least one of the inertial sensor in the array of inertial sensors; and
the processor determines the rotational rate of the movable member by solving for an output of the HMM using the input received from the array of inertial sensors, and controlling motion of the movable member using the rotational rate of the movable member.

6. The apparatus of claim 5 , wherein a state space size is constant as the number of inertial sensors of the array of inertial sensors increases.

7. The apparatus of claim 5 , wherein a pre-filter is used to approximate the rotational rate of the movable member, and the approximate rotational rate is used to discretize a state space.

8. An apparatus comprising:
a platform, wherein the platform comprises
an array of inertial sensors coupled to the platform, and
a processor coupled to the platform that receives inputs from each inertial sensors of the array of inertial sensors; and

a movable member coupled to the platform, wherein
each inertial sensor of the array of sensors outputs a rotational rate of the movable member,
the processor defines a hidden Markov model (“HMM”), wherein hidden states of the HMM represent a discrete value measurement of the rotational rate of the movable member, and transition probability of the HHM accounts for a motion model of the movable member, and observation probability accounts for noise and bias of at least one of the inertial sensor in the array of inertial sensors; and
the processor determines the rotational rate of the movable member by solving for an output of the HMM using the input received from the array of inertial sensors.

Assignees

• Univ Michigan Regents

Inventors

• Wang John
• Olson Edwin

Classifications

• G01C21/188

  for accumulated errors, e.g. by coupling inertial systems with absolute positioning systems · CPC title

• G01C21/166

  Mechanical, construction or arrangement details of inertial navigation systems · CPC title

• G01C19/58

  Turn-sensitive devices without moving masses · CPC title

• G01C19/00 Primary

  Gyroscopes; Turn-sensitive devices using vibrating masses; Turn-sensitive devices without moving masses; Measuring angular rate using gyroscopic effects · CPC title

• G01C19/5776 Primary

  Signal processing not specific to any of the devices covered by groups G01C19/5607 - G01C19/5719 · CPC title