Support Document

0 mins to read

An experimental Analytics Lab solution

Version-1 Lab Release: February 2021 Last document update: March 2021





Collision Detection

An Experiment in Analytics Lab

What is Collision Detection?

Collision Detection is a tool that reports potential collisions that are detected by Geotab’s telematics data within hours. This allows fleet safety managers to monitor, act, and take measures to mitigate risks. The Collision Detection tool is one of the many available tools through the Analytics Lab, found under Dashboard & Analytics in MyGeotab.

NOTE: Users on MyGeotab 2102 version can access Analytics Lab from the ‘Dashboard & Analytics’ menu in the left navigation bar.


Why do you need it?

While collisions are inevitable, incorporating the Collision Detection tool can reduce risk and encourage driver safety. Commercial fleets, in particular, have a lower tolerance for collisions due to higher financial and reputational risks, so being able to review and confirm potential collisions within hours is a useful protective measure.

The Collision Detection tool helps Fleet managers monitor potential collisions in their fleet vehicles by reviewing critical data points, taking action to reduce exposures, and mitigating risks of future events.


NOTE: This tool is independent of the Collision Reconstruction Add-in, which provides greater details on collisions.

How does it work?

The Collision Detection app leverages Geotab’s telematics data and reports all potential collision events in the user’s fleet database that show a high unusual acceleration reading in horizontal direction. A collision score is generated to indicate how likely the recorded acceleration event is a potential collision. An algorithm runs an analysis over each event and all events with scores 50% (previously, 60% based on the older model) or above are classified as a potential Collision, and the rest as No Collision.

Due to the granularity of this data, the algorithm can report collision details such as point(s) of impact, magnitude, trigger type (stationary vs. dynamic), location, time, and other metrics and statistics. The algorithm is based on the extracted insights from the data gathered through the recorded events. It supports factors like: acceleration and speed patterns, bad installation, event location, road type, as well as behaviour around an incident to estimate a collision score (in %).

These detected collisions are then ready to be reviewed by the user to confirm or deny their occurrences, and provide any additional information about the collision events. The app also provides an option to report an undetected collision event manually, which helps the algorithm improve future detections and reporting.

! IMPORTANT: Although all fleet accounts are eligible to use Collision Detection experiment, some Government and Public Sector fleets may be unable to access it. For more information, Resellers can contact their Partner Account Manager, and Customers can reach out to Support through their standard communication channel.


How to use the Collision Detection Experiment

Follow the steps below to navigate through the Collision Detection experiment and review, report, or monitor collisions.

NOTE: For demonstration purposes, this document includes screenshots of the experiment utilizing an internal Geotab database.


1

Install and launch the Add-in by navigating to Analytics Lab from the MyGeotab navigation bar.

NOTE: You must have Administrator security clearance to use the Collision Detection App.

2

On clicking the Try It icon on the catalog page, the landing page of the app will appear. By default, the App shows all collisions detected in the user’s fleet database.

3


To search for collisions, you can filter using the Show Events dropdown or filter based on specific locations using the Search Location dropdown.

4

Sort the data generated by Vehicle Name, Collision Score, Date, or Impact Points.

5

Select your vehicle to view further details on the collision detected. This provides a summary of:

  1. Impact points;
  2. Magnitude;
  3. Nature of collision;
  4. Acceleration and Speed profile charts; and
  5. User provided information.

6

To confirm the occurrence of the event, and to allow users to provide additional information, click Mark As Reviewed, and provide the required information. You can also click Archive to move and save the event.

7

Once the event is marked as reviewed, you can click Edit to change any responses to the event.

8

To report events manually, click Report a collision manually button on the landing page of the app.

Provide required details in the Manual collision report, and click Confirm.


Collision Detection API

A list of Collision Detection APIs can be found here. Please note that access to all Collision Detection APIs require users to review the terms and consent to the disclaimer form via the Collision Detection application in Analytics Lab, or the setUserConsent API listed in the document.

Learn More!

As we continue to better our Add-ins, your feedback is valued. Contact us with your feedback or questions by clicking Leave Feedback and completing our experiment feedback form.

If you have an idea for a new data experiment, click here and let us know!

Come join our Analytics Lab Community group to connect with our data experts, learn more about data inside Geotab and our upcoming experiments.

Version History

Version

Date

Owner

Updates

1.0

February 2021

Data Product Discovery Team

Initial Quick Guide

1.1

July 2021

Kriti Jangpangi

Updates based on the new model






scroll-up