Last mile delivery and quick commerce delivery is becoming available in more and more sectors. What used to be the Pizza delivery number has become a variety of apps that let you order food from all your favorite restaurants any time of the day. With covid-19 we also saw the rise of fast grocery deliveries with multiple providers now present in all big cities.
These services depend on localization technologies to identify the right driver to pick up an order at a restaurant, to notify the customers about estimated time of arrival and to manage and monitor their fleets efficiently.
The localization technologies are provided by the smartphones which are at the core of all of these services.
While delivery companies are thriving, unit economics are still tight. Many providers are still losing money on their fleets and market consolidations as well as investor expectations are raising the pressure. Considering the size of the fleets and the umber of rides, even slight adjustments of cost, up or down, have a significant impact on the operator’s bottom line.
The services depend on the built-in GNSS technologies of smartphones. While convenient, the technology has its downsides. Low frequency, significant errors especially in urban environments, high energy consumptions and missing information throughout drop off and pick up are some of the main drawbacks that limit further improvements of the customer experience and the service efficiency.
Using improved location and motion tracking solutions lets providers locate their riders more accurately, enables better insights on the different segments of the delivery process while reducing battery consumption of the riders’ mobile phones. This allows providers to improve their service efficiency and to improve the customer experience.
truemetrics offers precise location and motion data as a service. Our customers can implement our technology into their services with a simple SDK or using our API. Based on the data sent by their users’ devices we provide them with accurate location and motion data and offer the detection of motion patterns of interest, e.g. a rider getting off his bike to hand over the meal to the customer at their door.
This happens in three simple steps
While GPS relies solely on the quality of the data from the satellite navigation system we also use the data of all the other motion sensors which are available in modern smartphones. Accelerometers record the device’s acceleration, gyroscopes measure its rotation, barometers track the air pressure and magnetometers sense changes in the magnetic field. Data from all these sensors is recorded and sent to the backend via a plug and play SDK (or via an API endpoint).
Additionally, if available, data from IOT devices within the vehicles can be used to further improve the accuracy of the estimation.
The data is processed by truemetrics’ algorithm engine. It consists of two core components.
Sensor fusion and filtering for state estimation: We use the data of multiple sensors which track the devices motion, each type of sensor has its own advantages and disadvantages. By fusing the data from different sensors we can improve the overall accuracy, reliability and frequency of the estimated state of the device which consists of position, orientation, velocity and acceleration.
Pattern recognition with machine learning: The original sensor data as well as the results of the sensor fusion are the input for machine learning models which detect motion patterns of interest. These motion patterns are linked to events which are of interest to our customers for their specific use case. Examples for such events in delivery use cases are a rider getting of their bike for the final drop off or reckless driving.
The result data provided by the truemetrics is twofold: