System and method for predicting and responding to soft underfoot conditions

Patent metadata

Field	Value
Publication number	US-10288166-B2
Application number	US-201715657627-A
Country	US
Kind code	B2
Filing date	Jul 24, 2017
Priority date	Jul 24, 2017
Publication date	May 14, 2019
Grant date	May 14, 2019

Abstract

A soft underfoot conditions response system for use with a vehicle includes a plurality of sensors configured to transmit signals indicative of live data representing at least one of real time vehicle speed, vehicle acceleration, vehicle pose, vehicle payload, engine torque, engine power output, and engine RPM, and a controller communicatively coupled with the sensors. The controller is programmed to receive the live data, receive reference data representative of soft underfoot conditions from a database, and analyze the live data and the reference data. The controller determines a first set of parameters including measured real time values corresponding to wheel slip ratio and rolling resistance, vehicle speed, and vehicle pose, extracts from the reference data at least one of a first data subset containing vehicle operational parameters identified by an operator as being associated with soft underfoot conditions, and a second data subset containing data extracted using heuristics, and builds and trains a model for use by a classifier that segregates data subsets from the first set of parameters into a first classification that includes parameters that characterize surfaces with soft underfoot conditions, and a second classification that includes parameters that characterize surfaces without soft underfoot conditions. The controller also generates control command signals that cause a change in vehicle operational parameters to reduce or avoid any effects on operation of the vehicle associated with soft underfoot conditions.

First claim

What is claimed is:

1. A soft underfoot conditions response system for use with a vehicle, the system comprising:
a sensing system configured to generate signals indicative of live data representing at least one of real time wheel slip ratio and real time rolling resistance for the vehicle operating at a job site;
a plurality of sensors configured to transmit signals indicative of live data representing at least one of an image of a job site surface having features and characteristics associated with a presence or a lack of soft underfoot conditions, real time vehicle speed, vehicle acceleration, vehicle pose, vehicle payload, engine torque, engine power output, and engine RPM; and
a controller comprising one or more processors, and one or more memory devices, the controller being communicatively coupled with the sensing system and the sensors, and configured and programmed to:
receive the live data from the sensing system and the sensors;
receive reference data representative of soft underfoot conditions from a database;
analyze the live data received from the sensing system and the sensors, and the reference data received from the database;
determine a first set of parameters including measured real time values corresponding to at least one of an image of job site surface conditions, wheel slip ratio, rolling resistance, vehicle speed, and vehicle pose;
extract from the reference data at least one of a first data subset containing at least one of an image of job site surface conditions and vehicle operational parameters identified by an operator as being associated with soft underfoot conditions, and a second data subset containing data extracted using heuristics including identification of values for a mean rolling resistance greater than a threshold level of rolling resistance for greater than a first threshold time period, values for a vehicle speed greater than a threshold level of vehicle speed, and values for a wheel slip ratio greater than a threshold level of wheel slip ratio for greater than a second threshold time period;
build and train a model for use by a classifier that segregates data subsets from the first set of parameters into a first classification that includes parameters that characterize surfaces with soft underfoot conditions, and a second classification that includes parameters that characterize surfaces without soft underfoot conditions; and
generate control command signals that cause a change in vehicle operational parameters to reduce or avoid any effects on operation of the vehicle associated with soft underfoot conditions.

2. The soft underfoot condition response system of claim 1 , wherein the controller is configured to calculate the second data subset of the reference data using a sliding window approach, wherein a size of the sliding window depends on the data.

3. The soft underfoot condition response system of claim 2 , wherein the controller is configured to estimate features that include at least one of a mean of values of rolling resistance, a mean of values of wheel slip ratio, a mean of values of vehicle velocity, a variance in values of vehicle roll, a variance in values of vehicle pitch, and a variance in values of vehicle yaw for every second of data extracted from the reference data.

4. The soft underfoot condition response system of claim 3 , wherein the controller is further configured to normalize each feature dimension in the second data subset by subtracting each value for each feature dimension from a mean of all of the values for the feature dimension and dividing the result by twice a standard deviation for all of the values for the feature dimension.

5. The soft underfoot condition response system of claim 1 , wherein the controller is configured to build and train a model for use by a machine learning classifier using a K-Nearest Neighbor (KNN) search algorithm.

6. The soft underfoot condition response system of claim 5 , wherein the controller is configured to run the KNN search algorithm using a 10-fold cross-validation for determining a number of neighbors that results in an increase in accuracy of the model when using a Euclidean distance similarity metric.

7. The soft underfoot condition response system of claim 1 , wherein the controller is configured to build and train a model for use by a machine learning classifier using a Support Vector Machine (SVM).

8. The soft underfoot condition response system of claim 7 , wherein the SVM model is configured to classify measured real time values from the first set of parameters as characterizing machine operational parameters indicative of one of the presence of soft underfoot conditions or the lack of soft underfoot conditions.

9. The soft underfoot condition response system of claim 1 , wherein the controller is configured to determine the rolling resistance for the vehicle using values for engine torque, engine power output, and engine RPM derived from the plurality of sensors.

10. A method for predicting and responding to soft underfoot conditions, comprising:
transmitting signals from sensors associated with the vehicle and indicative of live data representing at least one of an image of a job site surface having features and characteristics associated with a presence or a lack of soft underfoot conditions, real time vehicle speed, vehicle acceleration, vehicle pose, vehicle payload, engine torque, engine power output, and engine RPM;
generating signals indicative of live data representing at least one of real time wheel slip ratio, real time rolling resistance for the vehicle operating at a job site; and
receiving the live data at a controller;
receiving, at the controller, reference data representative of soft underfoot conditions from a database;
analyzing, using a processor of the controller, the live data and the reference data;
determining a first set of parameters including measured real time values corresponding to at least one of an image of job site surface conditions, wheel slip ratio, rolling resistance, vehicle speed, and vehicle pose;
extracting from the reference data at least one of a first data subset containing at least one of an image of job site surface conditions and vehicle operational parameters identified by an operator as being associated with soft underfoot conditions, and a second data subset containing data extracted using heuristics including identifying values for a mean rolling resistance greater than a threshold level of rolling resistance for greater than a first threshold time period, values for a vehicle speed greater than a threshold level of vehicle speed, and values for a wheel slip ratio greater than a threshold level of wheel slip ratio for greater than a second threshold time period;
building and training a model for use by a classifier that segregates data subsets from the first set of parameters into a first classification that includes parameters that characterize surfaces with soft underfoot conditions, and a second classification that includes parameters that characterize surfaces without soft underfoot conditions; and
generating control command signals that cause a change in vehicle operational parameters to reduce or avoid any effects on operation of the vehicle associated with soft underfoot conditions.

11. The method of claim 10 , further including calculating the second data subset of the reference data using a sliding window approach, wherein a size of the sliding window is set to one second.

12. The method of claim 11 , further including estimating features that include at least one of a mean of values of rolling resistance, a mean of values of wheel slip ratio, a mean of values of vehicle velocity, a variance

Assignees

• Caterpillar Inc

Inventors

• Chrungoo Addwiteey
• Gowda Sandesh
• Kini Ananth Purushotham
• Wallstedt Philip Carl
• Davis Kevin Anthony

Classifications

• B60R21/013

  including means for detecting collisions, impending collisions or roll-over · CPC title

• G01C21/3461

  Preferred or disfavoured areas, e.g. dangerous zones, toll or emission zones, intersections, manoeuvre types or segments such as motorways, toll roads or ferries · CPC title

• G08G1/096791

  where the origin of the information is another vehicle · CPC title

• G08G1/162

  event-triggered · CPC title

• G08G1/096708

  where the received information might be used to generate an automatic action on the vehicle control · CPC title