US 12,351,185 B2
Method of determining parameters of interaction between a vehicle and a road
Gahl Berkooz, Ann Arbor, MI (US); David Wilson, Milford, MI (US); Pulkit Monga, Westland, MI (US); Grigory Yezersky, Farmington Hills, MI (US); Lev Levin, Charlotte, NC (US); Olivier Barree, Elancourt (FR); Michael Klank, Osnabrueck (DE); and Dennis Pape, Ostercappeln (DE)
Assigned to ZF Friedrichshafen AG, Friedrichshafen (DE)
Filed by ZF Friedrichshafen AG, Friedrichshafen (DE)
Filed on Oct. 17, 2022, as Appl. No. 17/967,197.
Claims priority of provisional application 63/256,836, filed on Oct. 18, 2021.
Prior Publication US 2023/0123958 A1, Apr. 20, 2023
Int. Cl. B60W 40/06 (2012.01); B60W 40/13 (2012.01)
CPC B60W 40/06 (2013.01) [B60W 40/13 (2013.01); B60W 2422/00 (2013.01); B60W 2556/45 (2020.02)] 9 Claims
OG exemplary drawing
 
1. A method of training a road condition monitoring model, the method comprising:
detecting, using one or more sensors on a vehicle, road surface data, under a controlled data collection operation during operation of the vehicle,
wherein the detecting step is accomplished by performing a vehicle operation over a first road surface under controlled conditions and executing the controlled data collection from the one or more sensors under the controlled conditions;
transmitting the collected data from the one or more sensors from the vehicle operation under the controlled conditions to a processing unit to operatively connected to the vehicle to create a raw reference data set from the data collected under the controlled conditions;
generating, by the processing unit a data pre-processing operation on the raw reference data set to create a training data set;
storing the training data set within the processing unit and
performing a data analysis of the training data set by the processing unit to determine frequency ranges of a section of the first road surface under a first predetermined condition to determine an absolute running mean;
wherein the processing unit is configured to generate an indication of abnormalities in the first road surface in remaining sections of the road surface by comparing the absolute running mean with the training data set; and
classifying abnormalities to create classifications rules within the training data set to define a preliminary model.