In issactoast/ikhyd: "I Know How you Drive - Analysis Tools for Telematics Data"

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

I Know How You Drive (ikhyd)

R package ikhyd is designed to make the driving behavior analysis using telematics data easier.

Prerequisite

Package installation and loading

To reproduce the research result, the \textbf{R package} ikhyd should be installed from github page, \url{www.github.com/issactoast}, using the following \textbf{R} code:

# packages loading
devtools::install_github("issactoast/ikhyd")

After the installation, load package as follows:

library(ikhyd)

Google API

To use the visualization functions, the google map aip key should be fed into the function as an argument. Google API can be obtained the Google Maps Platform by following the registration instruction. The google map API service will charge $2 per 1000 requests however, \$200 deposite will be added to the account per month. Thus, the personal usage is technically free but you need to register the credit card to use the service.

your_google_api <- keyring::key_get("GOOGLEMAP_API")

For example, track_vis() has the following structure:

track_vis(sample_trip$gps_data,
          api = your_google_api,
          zoom = 1)

How to access to data set

ikhyd package contains seven telematics files;

1. Sample route telematics: sample_trip.csv
2. Test route with OBD information:
   • trip_with_obd.csv
   • trip_with_obd.txt
3. Driving telematics files:
   • driver1.csv
   • driver2.csv
   • driver3.csv
   • driver4.csv

Note that trip_with_obd.txt contains the OBD speed information which combines with telematics data generated by smartphone, trip_with_obd.csv.

To get the path of the each files, users can use system.file() in R. For example, the follow code will give you the path of sample_trip.csv;

system.file("extdata", "sample_trip.csv", package = "ikhyd")

See help and actual code

You can check out the actual code in this instruction by typing the name of the R console. For example, examination the code of load_telematic_data() will be as follows:

load_telematic_data

Also, the help page of each function is available via ?function_name().

?load_telematic_data()

Load trip data

When you have a path of telematics data, you can load the data using get_trip() with data_option arguments as follows:

# telematics file path
sample_trip_path <- system.file("extdata", "sample_trip.csv", package = "ikhyd")

# load data
sample_trip <- load_telematic_data(sample_trip_path, all_in_one = TRUE)
summary(sample_trip)

Visualization

GPS data

The snippet of gps_data for the sample trip looks as follows:

head(sample_trip$gps_data)

The GPS coordinates of the sample trip can be visualized with track_vis() in ikhyd package as in the following Figure:

track_vis(sample_trip$gps_data,
          api = your_google_api,
          zoom = 1)

The following figure shows the Speed information from GPS sensor stored in speed_data, which can be visualized with plot_speed() as follows:

plot_speed(sample_trip$speed_data, tripname = "the sample trip")

Accelerometer data

For accelerometer information, you can plot it by using plot_acc() function with an option of smoothing parameter rate. Also note that some base plot options such as xlim can be used.

The figure below is generated by the following code.

plot_acc(sample_trip$acc_data,
         rate = 0.2, 
         tripname = "the sample trip")

Calibration

Kalman filtering and smoothing operation of the sample trip can be reproduced by the following functions: kalmanfilter_telematics() and kalmansmooth_telematics().

The visualization of the Kalman filtered and smoothed telematics data can be done by the plot_telematics() function in the package.

# telematics file for longer trips
obdtrip_path <- system.file("extdata", "trip_with_obd.csv", package = "ikhyd")
obdtrip_path_obd <- system.file("extdata", "trip_with_obd.txt", package = "ikhyd")

Kalman filter

kalmanfilter_result <- kalmanfilter_telematics(obdtrip_path)
head(kalmanfilter_result)

plot_telematics(kalmanfilter_result)

Kalman smoothing

kalmansmooth_result <- kalmansmooth_telematics(obdtrip_path)
head(kalmansmooth_result)

plot_telematics(kalmansmooth_result)

Comparison with OBD data

# load OBD trip data for the comparison 
obd_trip <- load_telematic_data(obdtrip_path, 
                                all_in_one = TRUE)

# load obd information and calculate acceleration
speed_data_obd <- get_obd_trip(obdtrip_path_obd)
speed_data_obd <- acc_from_obd(speed_data_obd)

Root Mean Square Error (RMSE) of accelerometer (y-axis), Kalman filtered accelerations, and Kalman smoothed accelerations for the given trip are calculated as follows:

rmse <- function(x, y){
    sqrt(sum((x - y)^2))
}

RMSEresult <- data.frame(
    Methods = "RMSE",
    Accelerometer = rmse(obd_trip$acc_data$y, speed_data_obd$dv_dt),
    KalmanFilter = rmse(kalmanfilter_result$a_lon, speed_data_obd$dv_dt),
    KalmanSmooth = rmse(kalmansmooth_result$a_lon, speed_data_obd$dv_dt)
)
knitr::kable(RMSEresult)

OBD based speed and accelaration

Speed comparison: OBD vs. GPS

obd_trip$speed_data$obd_speed <- speed_data_obd$speed
plot_speed(obd_trip$speed_data, col = "red",
           tripname = "OBD trip")

Acceleration comparison: Accelerometer vs. OBD

plot_data <- data.frame(time = obd_trip$acc_data$time,
                        acc1 = obd_trip$acc_data$y,
                        acc2 = speed_data_obd$dv_dt)
plot_acc_compare(plot_data, 
                 sensor_name = c("Y-axis accelerometer(red)", "OBD(black)"),
                 xlim = c(0, 400))

Acceleration comparison: Kalman filtering vs. OBD

plot_data$acc1 <- kalmanfilter_result$a_lon
plot_acc_compare(plot_data, 
                 sensor_name = c("Kalman filtering(red)", "OBD(black)"),
                 xlim = c(0, 400))

Acceleration comparison: Kalman smoothing vs. OBD

plot_data$acc1 <- kalmansmooth_result$a_lon
plot_acc_compare(plot_data, 
                 sensor_name = c("Kalman smooth(red)", "OBD(black)"),
                 xlim = c(0, 400))

Heatmap visualization

Here is some example of visualization of telematics data; v-a heatmap suggested by Wuthrich (2017) and Lee and Shyamal (2019) the Lon-Lat plot suggested by The following code generates the v-a heatmap of the telematics data

# telematics file for driver 1 and 2
driver1_path <- system.file("extdata", "driver1.csv", package = "ikhyd")
driver2_path <- system.file("extdata", "driver2.csv", package = "ikhyd")

telematics_driver1 <- kalmansmooth_telematics(driver1_path)
telematics_driver2 <- kalmansmooth_telematics(driver2_path)

Heatmap (V-A)

draw_vaHeatmap(telematics_driver1)
draw_vaHeatmap(telematics_driver2)

Heatmap (Lon-Lat)

par(mar = c(4, 4, .1, .1))
drawHeatmap(telematics_driver1)
drawHeatmap(telematics_driver2)

License

Provided under the terms of the MIT License.

Copyright © 2019-2020, Issac Lee.

issactoast/ikhyd documentation built on March 5, 2021, 10:24 p.m.