How to overcome to make mistakes to analyze a lot of sensors data which is generated in a driveless car. To make a prototype and extend it to a real situation will be great. The prototype application will be many like road condition monitoring, the car condition when a crash happens. All collected sensors data from the rover(mobile) can be published into AWS IOT and S3, through Kinesis and analysis of the data with multiple filtering combination - how much car was moving forward fast, what was the road condition based on 'accelerometer', 'gyroscope', 'magnetometer', 'light'. This mobile sensor data collector gives such application in the prototype level.
How I built it
Built Rover which has four wheels, and installed Raspberry PI 3 B, attaching XDK sensors kit, and Servo motors, and Sonar sensor. There are battery packs for Motor driving and Raspberry PI. All are integrated and commands are controlled by Linux shell command using Python.
Challenges I ran into
Integration with multiples sensors and motor controls are not simple and combines all data to send to AWS IOT. And the AWS IOT was simple to use, but it was challenged to integrate the data into Kinesis and Cloud Watch.
Accomplishments that I'm proud of
My Son and I worked together, that is most proud of thing. And AWS IOT experience will be a great opportunity to consider building up a real environment system.
What I learned
A lot of sensor data IOT area in real time can be hard for the human to filter and analyze without software help. In that sense, AWS IOT, CloudWatch, and Kinesis provide a man to set up IOT system very quickly. AWS IOT makes us handle to connect IOT devices very simply and easily, and could collect data over popular MQTT protocol. And CloudWatch provides debugging function when I did not see the IOT stream to Kinesis. Kinesis provides a very extensive filtering method to filter data using SQL.
What's next for Mobile Sensor Data Collector
Want to integrate the data into other AWS component like Dashboard , QuickSight. And study more how the mobile Sensor Data Collector can be used for future driverless car which will use sensors we used in this project. Put this prototype in crooked road, and collision test in different weather conditions.