IBM Asset & Facilities Management Connected Vehicle Insights - Public

IBM Asset & Facilities Management Connected Vehicle Insights - Public

Come for answers, stay for best practices. All we're missing is you.

 View Only

Analytics & Cognitive Feature in CVI Series: Part-2 Driving Behavior Analysis

By SHOICHIRO WATANABE posted Tue October 06, 2020 10:00 AM

  

Analytics & Cognitive Feature in CVI Series: Part-2 Driving Behavior Analysis

Overview

  • Insights into driving behavior can help automotive manufacturers to optimize design and production, manage quality, improve safety, and simplify maintenance. Vehicle fleet operations management and insurance companies can get powerful insights into customer vehicle usage and risk assessment.
  • You can analyze the behavior of drivers by using the data that is collected from the connected vehicles.
  • You can identify risky driving behavior and discover the actual driving patterns of a vehicle operator to identify unsafe practices or policy violations. For example, you can detect many different types of driving behavior or events out-of-the-box as product default plug-ins
  • In addition to the out-of-the-box, you can configure and customize to analyze and discover different driving behavior patterns of a vehicle operator and to get the insights that you need based on your project's requirements

Architecture of Driving Behavior Analysis

drb_overview


Data Flow

drb_dataflow



  • Data storage phase
    • VDH receives raw probe data via various protocol (REST API, MQTT, UDP, etc...)
      • Raw probe data is sent to DMM to execute map-matching with position information. And the incoming probe data is enriched with matched map data (static data: e.g. road type) and dynamic context data (e.g. traffic flow data, weather data) so that they can be referred in the later analysis phase
    • VDH stores enriched probe data into HBase
  • Data analysis phase
    • Driving Behavior Plug-in Framework receives a request to execute driving behavior analysis via REST API
      • Framework scans historical enriched data from HBase (Probe Historical Data)
      • Framework extracts behaviors, contexts and metrics based on configuration file and Java plug-in logic
      • (Optional) Framework stores the extracted behaviors, contexts and metrics into HBase (Driving Behavior Result)
  • See Online analysis of driver behavior data (Restricted content for entitled IBM customers)

Plug-in Deployment Flow: Customization


Out-of-the-box: Behaviors, Contexts and Features Extracted By Default Plug-ins

Behaviors

Driving behavior type Description
Harsh acceleration Harsh acceleration threshold value in accordance with the speed is internally set. Harsh acceleration is recognized if acceleration data exceeds the threshold value.
Harsh braking Harsh deceleration threshold value in accordance with the speed is internally set. Harsh braking is recognized if the braking data for a trip exceeds the threshold value.
Speeding Over speed threshold value per road type is set. Speeding is recognized if the speed data for the trip exceeds the threshold value. You can configure the threshold for speeding.
Frequent stopping A stop is detected when the speed value is zero. Frequent stops are recognized when many stop instances within a certain time frame in the trip are detected.
Frequent acceleration Acceleration threshold value in accordance with the speed is internally set. Frequent acceleration is detected if the driver frequently exceeds the threshold value within a certain time frame during the trip.
Frequent braking Deceleration threshold value in accordance with the speed is internally set. Frequent deceleration is recognized if the driver exceeds the threshold value frequently within a certain time frame during the trip.
Sharp turn (high-speed turn) Speed threshold value in turn is set. A sharp turn is recognized if the driver exceeds the threshold value. You can configure the sharp turn threshold.
Acceleration before turn Acceleration threshold value in accordance with the speed is internally set. Acceleration before a turn is recognized if the driver exceeds the threshold value.
Over-braking before the vehicle exits a turn Deceleration threshold value in accordance with the speed is internally set. Over-braking before a vehicle exits a turn is recognized when the driver exceeds the threshold value.
Fatigued driving Fatigued driving threshold is set. Fatigue in a driver is recognized if the driving duration for a single trip exceeds the threshold value.

Contexts

Context category Description
Time range Day, Night, Morning peak hours, Evening peak hours
Road type Motorway, Urban highway, Urban primary, Urban road, Others, Unknown
Speed pattern Free flow, Steady flow, Severe congestion, Congestion, Mixed conditions

Features

Feature Description
distance Distance (mileage) of a trip
time_span Duration of a trip
average_speed Average speed by dividing trip distance by trip span
max_speed Maximum speed
idle_time Idle duration of a trip
night_driving_time Duration of driving in night time
rush_hour_driving Duration of driving in evening rush hour
month_of_year Month of year
day_of_month Day of month
day_of_week Day of week

Example of Extracted Behaviors From Starter App


  • This screenshot is an example of extracted behaviors which comes from starter application
  • The Fleet Management starter app is available to showcase the connected car features of CVI
  • In this example, the extracted behaviors are shown in left upper side and the extracted features are displayed on right lower side.
    • A trajectory is displayed as blue line and the red overlapped dots show the extracted behaviors in the trajectory
extracted_behaviors

Summary

  • Out-of-the-box default product plug-ins are supported to detect some driving behaviors, contexts and features
  • Default product plug-ins are customizable by configuring the threshold values
  • Driving behavior framework is designed to support customizability for project specific requirements. You can develop custom logics as additional plug-ins to incorporate project specific requirements

References

Notice

  • Information in this blog is based on CVI SaaS V3.1. Information is subject to change in the future.

Revision History

  • 2020-10-06: Initial release

#Sustainability
#IBMAssetandFacilitiesManagementConnectedVehicleInsights-Public
0 comments
20 views

Permalink

https://community.ibm.com/community/user/blogs/shoichiro-watanabe1/2020/10/06/analytics-cognitive-feature-in-cvi-series-part-2