Mobility Analytics

Custom Data Science Services For The Automotive Ecosystem

ABC car-sharing, car leasing, OEM, tire manufacturer, car dealership, or any other automotive-related company wants to:

  • Create new data-driven products, and/or
  • Solve challenging data objectives, and/or
  • Understand what they can accomplish with their available data, and/or
  • Get analytics to work on research projects in-house

Mobility Analytics: Our Approach

Motion-S can deliver intelligent insights based on business and data understanding thanks to its powerful mobility data processing platform and specialized data science team. We help you to discover patterns in your data and build innovative solutions. Our approach is simple:

Together with our customer, we define the problem to solve and the data-enabled product features. Next, we do a first exploration of collected data (existing or collected via our SDK) and propose a way forward in terms of data modeling. With that in mind, we prepare the datasets and focus on the modeling aspects, by testing different AI and ML approaches and evaluating performance. Last, we prepare the deployment into the customer environment and the integration into their environment.

1654

Let's dive into a few use cases:

Predictive Maintenance for Heavy-Duty Vehicles

Automotive company ABC is looking to build an innovative data-driven product, and needs help to develop predictive models, explore and take conclusions on data collection alternatives, augment their trip data with contextual information, and understand the optimization impact of their data-driven models. Motion-S can help. Let's take the following example.

Monitoring wear of heavy-duty vehicles is standard. But how about a predictive approach, limiting downtimes and optimizing maintenance interventions? Data science offers a fantastic mean to develop predictive maintenance models based on a representative fleet.

Our approach in this case:

  • Stage 1: Our data science team conducts research in the topic to identify different predictive models for vehicle maintenance.
  • Stage 2: The team tests different data collection options to understand data fields availability.
  • Stage 3: Data is collected from vehicles, including: GPS location of trips, data from accelerometer, engine data from OBD, and diagnostic trouble codes (DTCs)
  • Stage 4: Trip data is processed in the Motion-S platform, adding contextual information to the trips.
  • Stage 5: A Predictive maintenance model is developed, using DTCs, engine operating data, and contextual scenarios.
  • Stage 6: Predictive model is promoted to the platform and integrated via API to Client Fleet Management system

You are keen in doing a similar project? Don't hesitate and contact us today.

Supporting Research Projects with Accurate Augmented Data

To augment and profile trip data, many companies rely on the Analytics endpoints offered by the Motion-S platform. These endpoints offer a plug & experience to obtain contextual and behavioral information, and can be customized or expanded based on the client needs. To illustrate this better, let's take the following example.

An automotive component manufacturer is interested in building models that explain and predict how the context of driving wears down the performance and condition of the vehicle components. For example, factors such as slope, curvature, road type, roughness of the road, pavement type, temperature, precipitation, and relative humidity are important to understand how fast tires deteriorate.

In order to obtain this data, the automotive component manufacturer sends trips to the Motion-S platform and gets back the augmented trips including details per location of the context (weather, road topology, road conditions). In addition, Motion-S creates new endpoints that can help the client in building their models, be it by expanding the available contextual information with new data layers, reconstructing historical trips or offering advice to test new data hypotheses.

Another use case, is the identification of a low-emission driving style that can be fostered through driver training, a driver support app, and guidelines for regular maintenance to ensure efficient, resource-saving conduct on the roads.

The available and collected driving and location data via a mobile application is enriched with context to gain insights into the factors influencing low-emission driving. Sending the generated raw GPS data to the platform allows creating a data lake. A set of callbacks for getting the trip-based raw (JSON) augmented and sending profiled data back to a set of specified endpoints is created. During the augmentation phase, more than 100 different information elements are added to each location and road segment of the trip, e.g., road characteristics and quality, road topology and limitations, traffic signals, and weather information, to name only some.

Get in touch to learn more!

Providing mobility analytics to a car-sharing company

Car-sharing is gaining more and more popularity. Especially in larger cities and metropolitan areas, not everyone needs to own a car as public transportation is available, when and where it matters.
Car-sharing operators need to take care of the operational state of their fleets vehicles. Only knowing where the cars are located is insufficient, but maintenance needs have to be transparent to the fleet manager. Understanding how a customer drives a car provides deep insights into accident risk or possible damage.

2670

Collecting And Submitting Data

The car-sharing company decides to deploy OBD2-based data collection devices in the fleet's vehicles. The collected data, in principle just a succession of bytes, is sent via a custom solution to get the raw data from Teltonika and submit and interpret this data.

Data Obtained

The data obtained from the OBD2-based data collection devices is providing information on:

  • Engine Oil temperature
  • Fuel Level
  • Acceleration X,Y,Z axis
  • Engine Runtime
  • Distance Since Codes Cleared
  • Absolute Fuel Air Pressure
  • Control Module Voltage
  • Absolute Load Value
  • Absolute Fuel Air Pressure
  • Fuel Injection Timing
  • VIN Number
  • Hybrid Battery Pack Remaining Life
  • Number of DTC
  • Fuel Pressure
  • Throttle Position
  • Angle
  • Vehicle Speed
  • Latitude
  • Longitude
  • Altitude
  • Ignition
  • Movement
  • Engine Load
  • Coolant temperature
  • Barometric Pressure
  • Intake MAP
  • Intake Air Temperature
  • Commanded EGR
  • EGR Error
  • Time run with MIL on

πŸ“˜

Good to know:

Which data can be collected from the device depends on the vehicle.

Sample Payload

{
            "MAF": 516,
            "VIN_number": "WBA7J320301E20124",
            "absolute_load_value": 24,
            "acc_x": -116,
            "acc_y": -307,
            "acc_z": 4,
            "altitude": 119,
            "ambient_air_temperature": 2,
            "angle": 74,
            "barrometric_pressure": 98,
            "control_module_voltage": 14800,
            "coolant_temp": 102,
            "distance_since_codes_cleared": -26060,
            "engine_load": 13,
            "engine_oil_temp": 102,
            "engine_rpm": 1485,
            "engine_runtime": 706,
            "fuel_level": 50,
            "ignition": 1,
            "intake_MAP": 91,
            "intake_air_temp": 24,
            "movement": 1,
            "speed": 28,
            "vehicle_speed": 36
        }

Keyfindings

Safety, Wear, and Eco-Efficiency

Analytics on Risk, Energy Efficiency, and Vehicle Wear can be obtained through the API. The car-sharing company can get insights into how well their customers drive and identify the worst drivers or the best to either coach them in the first case or reward them.

1006

New insurance model without the need of being backed by an insurance company

Car-sharing is flexible, but insurance premiums are not: they are usually fixed to a particular premium based on specific characteristics of a driver, e.g., age and years of driving experience. With the upcoming popularity of Pay-as-you-drive insurance products, taking into account, the distance traveled and offering more flexible pay as you go tariffs, the insurance business is changing rapidly.
With accurate information about the distance traveled in a certain period, let's say a month, the car-sharing company can fix a base price per kilometer and offer additional distance packages for insurance. Using Trips statistics returns all required information to offer a Pay-as-you-drive product.

curl --request GET \
     --url 'https://api.motion-s.com/analytics/info/v1/stats' \
     --header 'Accept: application/json'
     --header 'X-Api-Key: <Your API Key>'
[
  {
    "number_of_trips": 691,
    "total_distance": 11082,
    "number_of_locations": 627466,
    "number_of_drivers": 44,
    "number_of_devices": 52,
    "first_trip_date": "21-07-2020",
    "last_trip_date": "04-01-2021",
    "average_speed": 45,
    "total_time_travelled": 1508
  }
]
2609

Moving ahead, integrating safe driving behavior to offer even better insurance tariffs based on our risk assessment can reward good drivers and boost loyalty, and motivate other drivers to change their behavior. With Risk summary, the car-sharing company is able to retrieve the risk score for a given driver. Combining distance traveled and driving styles will lead to a competitive insurance offer, fair and transparent to the clients.

curl --request GET \
     --url 'https://api.motion-s.com/analytics/risk/v2/summary?days=14' \
     --header 'Accept: application/json'
     --header 'X-Api-Key: <Your API Key>'
{
  "score": 89,
  "subscores": {
    "speeding": 92,
    "complying": 86,
    "cornering": 89,
    "accelerating": 90,
    "braking": 86,
    "resting": 92
  },
  "factors": {
    "speeding": {
      "frequency": 0.07,
      "intensity": 1.05,
      "penalty": "low",
      "bad": 5141,
      "total": 77373
    },
    "complying": {
      "frequency": 3.29,
      "intensity": "low",
      "penalty": "low",
      "bad": 61,
      "total": 69044
    },
    "cornering": {
      "frequency": 0.04,
      "intensity": "low",
      "penalty": "low",
      "bad": 16,
      "total": 589
    },
    "accelerating": {
      "frequency": 0.04,
      "intensity": "low",
      "penalty": "low",
      "bad": 33,
      "total": 566
    },
    "braking": {
      "frequency": 0.05,
      "intensity": "low",
      "penalty": "low",
      "bad": 27,
      "total": 397
    },
    "resting": {
      "bad": 1,
      "total": 39
    }
  }
}

Vehicle Wear

Creating a vehicle wear profile is of added value as it helps the car-sharing company decide whether they should keep the car in their fleet or sell it.
Estimating vehicle wear can be done based on contextual data such as

  • The history of past routes, weather conditions, projected annual distance
  • Static information such as make and model, and
  • Vehicle operating conditions, DTCs, extreme IMU (near-to-crash or crash), obtained from OBD-II.

Predictive maintenance to avoid unexpected breakdowns or to better plan maintenance activities is possible by using

  • Digital odometer information to proactively plan service (take the appointment for the user, pick the place, reserve a replacement vehicle, etc.)
  • Digital DTC information to react on time to defects that may damage the vehicle in the mid-long term

Boosting good driving behavior can be achieved by rewarding drivers that expose the vehicle less often to extreme situations.

Using our Vehicle Wear API calls provide the company with all relevant information to better assess vehicle health.

curl --request GET \
     --url 'https://api.motion-s.com/analytics/wear/v1/summary?trip_id=-1&<DRIVER_ID>' \
     --header 'Accept: application/json'
     --header 'X-Api-Key: <Your API Key>'
[
  {
    "wear_score": 76.93,
    "battery": {
      "battery_score": 44.3,
      "charge_efficiency_score": 44.3
    },
    "body": {
      "body_score": 83.18,
      "pct_locations_with_dense_traffic_score": 75.67,
      "commercial_area_score": 90.68
    },
    "brakes": {
      "brakes_score": 77.22,
      "pct_locations_with_high_braking_score": 69.83,
      "braking_dowhhill_density_score": 84.61
    },
    "engine": {
      "engine_score": 90.83,
      "jackrabbit_start_density_score": 100,
      "temperature_in_degrees_score": 100,
      "pct_locations_with_inefficient_speed_score": 80.5,
      "pct_locations_with_harsh_accl_score": 73.68,
      "pct_locations_with_harsh_acc_at_low_temp_score": 100
    },
    "suspension": {
      "suspension_score": 75.09,
      "proportions_performace_according_to_IRI_score": 75.09
    },
    "transmission": {
      "transmission_score": 60.88,
      "gear_change_density_score": 60.88
    },
    "tyres": {
      "tyres_score": 74.05,
      "proportions_road_quality_score": 47.78,
      "proportions_temperature_score": 98.52
    }
  }
]