Granted Patent B2

US 9,984,513 · App. 14/582,118 · Granted May 29, 2018

Inventors: Anurag Ganguli (Milpitas, CA); Rajinderjeet Singh Minhas (Mountain View, CA); Johan de Kleer (Los Altos, CA)

Assignee: Palo Alto Resarch Center Incorporated

G07C5/0808G06Q10/20G06Q30/014G06Q30/018G07C5/006G06Q50/30

Abstract

Claims (23)

1. A method for determining vehicle component conditions, comprising:

storing diagnostic functions that map one or more features of motor current measurements to different door conditions, wherein each door condition comprises a problem with operation of a vehicle door;

obtaining the motor current measurements for a door over a predetermined time period;

identifying at least one open and close cycle of the door during the predetermined time period comprising movement of the door from an open state to a closed state or a closed state to an open state;

selecting the motor current measurements occurring during the at least one open and close cycle;

calculating the features from the motor current measurements obtained during the at least one open and close cycle, comprising one of:

performing:

dividing the at least one open and close cycle into at least three different states, wherein each state is identified using a different predetermined rule based on one or more of motor current, door commands, and operating state;

identifying the motor current measurements for each state; and

calculating statistical measurements as the features for each state in the at least one open and close cycle; and

performing:

plotting the motor current measurements for the at least one open and close cycle on a graph;

fitting a mathematical function to the plotted motor current measurements for the at least one open and close cycle; and

identifying coefficients of the fitted mathematical function of the motor current measurements as the features;

analyzing the features of the motor current measurements by mapping the features to one or more door conditions via the stored diagnostic functions; and

identifying from the different door conditions at least one problem with the door based on the analyzed features that correspond with the door condition for that problem.

2. A method according to claim 1 , further comprising:

determining the diagnostic functions that map at least one of the motor current features to the different door conditions via supervised machine learning.

3. A method according to claim 1 , further comprising:

dividing the motor current measurements for a single open and close cycle of the door into states prior to determining the features of the motor current measurements, wherein the features of the motor current measurements include a subset of features for each state in the at least one open and close cycle of the door.

4. A method according to claim 1 , further comprising:

determining the states of the motor current measurements for the at least one open and close cycle of the door using a set of rules dependent on the motor current measurements and related to different phases of the at least one open and close cycle.

5. A method according to claim 1 , wherein the different door conditions each comprise one of presence of dirt, lack of grease, one or more loose bolts, a presence of objects foreign to the door, and rail deformity.

Assignments (10)

Continuity (1)