This code implements risk event detectors that categorize driving style.
The directory has the follopwing structure
- applications / insurance_event_detector contains the python code for the main application
- digital-twin-providers / vehicle_properties_provider contains the python code that registers the signals and simulates the vehicle, described as 2b in the diagram
- digital-twin-model contains a DTDL representation of the COVESA signal specification used in the project.
- proto-build contains all Python gRPC interfaces to the different components (service_discovery, invehicle_digital_twin, module)
Each Risk event detector monitors vehicle signals everytime that they change, at a given refresh rate. The algorithm of each event detector varies - in some cases it requires analyzing previous data, or it can be a simple ramp up or ramp down threshold detection.
Once an event is triggered, the event detector will capture additional signals before and after the event.
The following events react on a flank up or flank down
| Event | Monitored Signals | Additional Captured Signals |
|---|---|---|
| speeding_start | Vehicle Speed | |
| speeding_end | Vehicle Speed | |
| cruise_control_activated | ADAS_CruiseControl_IsActive | Vehicle Speed |
| cruise_control_deactivated | ADAS_CruiseControl_IsActive | Vehicle Speed |
| harsh_braking | Vehicle_Acceleration_Lateral | Vehicle Acceleration Longitudinal / Lateral, Speed, Brake Pressure, Steering Wheel Angle, ADAS ABS Error, ADAS ABS Is Engaged, ADAS TCS IsEngaged |
| harsh_acceleration | Vehicle_Acceleration_Lateral | Vehicle Acceleration Longitudinal / Lateral, Speed, Accelerator Pedal Position, Steering Wheel Angle, ADAS ABS Error, ADAS ABS Is Engaged, ADAS TCS IsEngaged |
| harsh_cornering | Vehicle_Acceleration_Longitudinal | Vehicle Acceleration Longitudinal / Lateral, Speed, Brake Pressure, Accelerator Pedal Position, Steering Wheel Angle, Steering Wheel Angle Sign, Left Turn Light Indicator, Right Turn Light Indicator, ADAS TCS IsEngaged. Vehicle Speed Wheel Front Left / Front Right / Rear Left / Rear Right |
The Insurance Event Detector uses charriot to discover the digital twin service. It will collect all necessary signals and use the digital twin service to detemrine if they exist. Once the signals are discovered, it will use the managed subscribed (through the digital twin service) to read the metadata and connect.
To provide simulation, a vehicle provider will register signals (described in a DTDL file) and provide updates using a sample CSV file. The CSV file has the following structure:
source_id, signal, timestamp, value
"b7ed5744-8715-4b3d-a322-7c7e0c399f69","Vehicle_Speed_Wheel_RearLeft","142.785732","53.715"
"b7ed5744-8715-4b3d-a322-7c7e0c399f69","Vehicle_Speed_Wheel_FrontRight","142.785732","54.3975"
"b7ed5744-8715-4b3d-a322-7c7e0c399f69","Vehicle_Speed_Wheel_FrontLeft","142.785732","53.685"
"b7ed5744-8715-4b3d-a322-7c7e0c399f69","Vehicle_Speed_Wheel_RearRight","142.785732","54.225"
As a prerequisite to run he application, it is necessary to run the base modules of the orchestration blueprint. Make sure that the network is the same (e.g. host).
The following modules should be running - for example, when running using Eclipse Ankaios using docker and the local network from the Eclupse Orchestration project (asumming the code is checked out in ~/repos/maestro-challenge).
docker run -it --privileged --name custom_ankaios_dev -v ~/repos/maestro-challenge/eclipse-ankaios:/workspaces/app -v ~/repos/maestro-challenge/in-vehicle-stack:/workspaces/app/in-vehicle-stack --network host -p 25551:25551 --workdir /workspaces/app custom-ankaios-dev:0.1 /bin/bashfollow the instructions to run the base workloads using the run_maestro.sh script. listing the workloads
/workspaces/app# ank get workloads
WORKLOAD NAME AGENT RUNTIME EXECUTION STATE
digital_twin_cloud_sync agent_A podman Running
digital_twin_vehicle agent_A podman Running
dynamic_topic_management agent_A podman Running
mqtt_broker agent_A podman Running
service_discovery agent_A podman RunningOnce this is running it is possible to run the vehicle_provider and the insurance_event_detector applications