In the example on this page, we present an example scenario where the user is asked to make a short drive. Throughout the scenario, the driver encounters various events (road crossing, pedestrian crossing) The simulation automatically checks if the driver took the appropriate action in the given situation, but does not provide this feedback to the user. Instead, this data is uploaded directly to the cloud, where it becomes immediately accessible for researchers.
The shared Google Colab (Jupyter Notebook) that is presented below, provides an example of how data can easily be accessed, shared and analyzed.
Demo Scenario
Run the demo scenario below (Only works on PC) by filling in a unique user code and clicking start. Use the arrow keys to drive/brake and look left/right. Pay attention to any events and drive as you would drive in real life. Once the end of the scenario is reached, data will automatically be uploaded.
Accessing the demo data
Data collected by the demo application can be accesses through the shared example jupyter notebook. The notebook has some example code to achieve fololwign functions:
- Access the full dataset (collected from all users)
- Visualize aggregated data
- Visualize individual data
- Download data files
1. Run All command
Use the Run all command from the top task bar to get the latest data from the server
2. Aggregated Data
Aggregated data is visualized and can be found by scrolling down in the notebook or jumping to the Aggregated data section from the table of contents.
The example code shows the current user count, average score and score per user and visualizes the score distribution as a bar-chart.
3. Individual data
Individual data for a specified user can also be visualized and can be found by scrolling down in the notebook or jumping to the Individual data section from the table of contents.
- Use the input form the fill in your user code (if you have performed the demo scenario)
- Use the Run All command from the top bar.
- As an example, the speed trace for the specified user is now visualized.
4. Download Data Files
Through the shared notebook, individual and aggregated data files can also be downloaded.
On the left-hand side taskbar, navigate to the file menu. Here, it is possible to view and download the automatically generated data files:
- An aggregated data file containing event data for all users.
- An event file for the specified user, containing information about how the user interacted with events in the scenario.
- A detailed data file for the specified user, containing data logged during each simulation frame.
Conclusion
By taking full advantage of Unity WEBGL technology, it is possible to easily deploy driving scenarios with simple tasks that do not require a full-size simulator on a large scale.
Storing datafiles directly to the cloud can be achieved with web-based scenarios, but also from a full size simulator experiment. It comes with several advantages over storing data locally:
- Automatic backups ensure data is never lost
- Data can be made available through an API.
- sharing and colaboration between researchers can easily be achieved. For instance with a shared Google Colab environment.
- At an early stage, preliminary results can easily be visualized and shared with stakeholders.
For more information about our web- and Unity-based solutions, send us a message on:
Recent Comments