Run the Test

Online Test

When ready to begin a test, press the test START button. Test Suite will begin logging, but the test will not start until you press the trigger to signify TGaze. TGaze is the time when the driver's eye finds the gaze location (camera 2 when using this methodology). This is manually defined by pressing the handheld trigger.  

This gives you time to prepare the vehicle, get to steady state and begin the test when ready by pressing the trigger and setting the first TGaze. Test Suite will capture data from when the Start button is pressed, which is important for aspects of various assessments, such as ‘4 seconds attentive driving’.  

The important measurables for this assessment are:  

• TGaze
• TAway  
• Gaze Transition (TGaze-TAway)  
• T0  
• Twarn (EoT condition given by the start of an Audio or Visual warning to the driver)
   

The driver should then keep focus on this camera point (Gaze position) until EoT which is either of the following: 

• Twarn (rising edge of Audio or Visual trigger) 

• Test timeout (defined within test configuration; default = 5 seconds)   

Following the EoT the driver can either go straight into another gaze position scenario, repeat the same gaze-position scenario or finish the test. To finish logging, press the STOP button.  

The run results will be shown in the table, including Pass/Fail, Test Duration and TAudio/TVisual, denoting the duration from the start of the test until the warning.  

You can now assess if any runs require post-processing as edge cases.  

For example, if a manual error of up to 100 ms is assumed, then run number 1 (10 ms from the Pass/Fail limit) and run 6 (40 ms from the Pass/Fail limit) could potentially be considered edge cases to go forward for post-processing. The other runs were either Pass or Fail or No reaction, and all outside of the 100 ms manual error:  

Example of online data collection

Offline Test

It is up to each user to consider what is defined as an outlier for the test. Nominally, the error margin for a skilled test engineer is in the region of 30 ms to 100 ms (30 ms is the difference between 2 frames, hence the minimum visual error margin detectable), therefore, it could be considered that any results that sit in the region of  ± 100 ms from a Pass/Fail result could be re-evaluated (as discussed in the section above).  

To perform the post-process assessment, you must first stop the logging on both HD2 units and remove the SD cards. The data on each SD card can be stored locally, and you can load it in a second session in Test Suite. This session must be in Offline mode, and you must load the data from both HD2 units.

Once both HD2 files have been loaded into Test Suite, they will automatically sync and move frame-by-frame as you navigate the files.  

The Results tab should show the same results as the Online test and the graph will show the exact start and end of the test (defined by the Trigger for test start and Audio/Visual trigger for test end).  

You can now identify the runs (tests) considered edge cases in your scenario. These edge cases can then be assigned for scrutinisation or further analysis refined by TGaze position in time. While the Online tests give a localisation of this position based on only the manual trigger which gives an appropriate accuracy for the majority of the tests, you can now refine those close (ca. 100 ms) to the fail criteria with the Add t_Gaze button:

1. Place the cursor on the start of test (Trigger on).
2. Use Ctrl + arrow to scroll frame by frame to define when the driver's eyes settled on the TGaze position.
3. If TGaze differs from the start of test defined by the manual trigger, you can assign a new modified test start by clicking the Add t_Gaze button to add a t_Gaze marker. 
4. You can repeat this process for any/all of the test runs if required.  

Example: 

When using the data in the example above, we can do an offline assessment to decide if the pass/fail result of the edge cases was affected by manual error.  

Run 1 was 0.01 s under the threshold, and Run 6 was 0.04 s over the threshold.  

When reviewing the files frame by frame, we could ascertain the absolute TGaze time and modify the manual trigger position in the data to match this absolute time.

In this example, the trigger was pressed before the eyes completed their movement to TGaze. By moving the cursor frame-by-frame, we could see the exact time when the eyes focused on the Gaze position and place a t_Gaze marker at that location instead. After clicking Refresh, the data was modified to reflect the new duration. 

We modified Run 1, and the total duration reduced by 0.03s (1 frame) and the run remained a Pass in the results.  

Run 6 did not require modification as the trigger was pressed in the same frame as the eyes focused on the TGaze position and remained as a confirmed Fail in the Results.  

We confidently confirmed the Absolute Pass/Fail results for the scenario being assessed and can now move to the next scenario or Gaze location.  

Gaze transition 

Gaze Transition is an important validation criterion for Euro NCAP assessment. The DSM test plugin has seek buttons for the Gaze Durtion, making it easy to navigate to the closest TGaze position.

There are 3 different time criteria for the 3 different movement types: 

• Lizard (0.16s) 

• Owl (0.48s) 

• Body Lean (1.0s) 

By clicking the corresponding time buttons in the Seek t-Gaze area, Test Suite will shift the data and video to the corresponding position before the TGaze position. You can then simply judge if the Gaze duration was within the required time when reviewing the video. 

-4.00s 

You can review the last 4 seconds before the start of the test to confirm the requirement of 4 seconds of attentive driving before TGaze.  Click the -4.00s button to move the cursor to 4 seconds before the start of the test.