eXpress
UI/UX Design
Data Engg & Analytics
Mobile App
A Logistics company with their fleet of trucks delivering goods across the country. The company wants to bring in a connected ecosystem and leverage AI to help their team build smarter systems. The company then hired a 3rd party solutions team to connect data from the telematic devices on the trucks to the cloud and capture the data. Here are few things they wanted to do with this data.
Industry
Logistic
Company size
<100+ Employee
What we did
- Data Engg & Analytics
- Mobile App Development
Objective
A Logistics company with their fleet of trucks delivering goods across the country. The company wants to bring in a connected ecosystem and leverage AI to help their team build smarter systems. The company then hired a 3rd party solutions team to connect data from the telematic devices on the trucks to the cloud and capture the data. Here are few things they wanted to do with this data.
Use the data to get a real-time view of their fleet including parameters such as GPS location, engine temperature, speed, acceleration and consignment parameters.
Bring in data from other systems to optimise the routes based on traffic conditions, weather and consignment size.
Build an alerting system to trigger in case the consignment diverted from the regular or recommended route.
Monitor vehicle vitals and health and provide predictive maintenance schedules.
Monitor driver behaviour and provide corrective recommendations in-real time
Challenges
Once the data from the devices were connected to the cloud. The team used AWS kine sis to stream the data into S3 and kine sis analytics where the team could analyses the stream and process it. The processed data was then pumped into DynamoDB where the real-time dashboards were connected. The team had challenges with the cloud native setup as mentioned here.
The data visualisations was restricted to AWS quicksight features
When new team members were added, complex IAM rules had to be applied to give them access to the right parts of the system.
Deploying Sagemaker model inferences on edge systems was very complicated and required assistance from 3rd party solutions team and AWS consultants
The overall system was complicated and had many internal dependencies making it rigid due to cloud vendor restrictions and also pricing models were complicated and expensive
Solution
Streaming Data sources
Synctactic supports all kinds of streaming data from kafka streams to MQTT. In this case the data was sent to the cloud using AWS kinesis and this was replaced by kafka for it’s flexibility. Once added to synctactic a simple search interface could be used to explore data in the stream. A project was created with this source. The pipe was used to execute the transformations on the stream and store them into redshift. This was then connected to a realtime dashboard on the synctactic platform on which the team could see real time data of the platform including custom mapbox route traces.
Route Optimisation
Real-time traffic data from 3rd party APIs was added via the API designer on synctactic. Weather data was also used as feature. This data was easily added and enriched the historic trip data. The data was processed and the features generated were fed to a proprietary algorithm to predict delivery time based on specific routes. This way the company could plan shipments better to hit their SLAs.
Alert Mechanisms
The recommended routes were fed to the truck driver via the company’s app. The routes are also shown on the synctactic dashboard and in real- time the team can monitor the live GPS location of each of their fleet and it’s recommended route to a given destination.
Predictive Maintenance
The sensors on the truck keep sending vital information on the health of the truck to synctactic. A project was created to analyse this data. Multiple data sources were combined to provide the inputs to the model that could predict a window during which the truck had to be serviced and to prevent breakdowns. This helped the company save a lot on maintenance costs as any problems with the vehicle could be assessed ahead of time.
Driver Assistance systems
The sensor data on the truck along with driver inputs were added as sources to synctactic. These data points were analysed and a model was trained to classify between good and bad driving behaviour. The output of this model was fed back to the driver to suggest speed to maintain, ideal acceleration during stalls and gear shift behaviour. This system helped the company improve driving as well as fuel efficiency of their trucks.
Outcome
With Arokee AI, the company was able to bring in a data science based approach to look after their operations and hence improve customer satisfaction. They could optimise routes and stick to SLAs and Reduce operational costs by equipping drivers with feedback systems. Arokee AI provided them a way to leverage a scalable and flexible data infrastructure for all their data science initiatives.
Improved team productivity
Recent Work
Launch your product with Arokee