Project Design
For the embedded system, we use four ground force sensors and a temperature/humidity sensors for data collecting, all the sensors are connected to an Arduino UNO board, which convert the analog signal to digital signal and the converted data is feed to a raspberry pi.
On the raspberry pi, we use a camera to track customers’ faces, and record the time they come and stay. The raspberry also process and upload the preliminary data;
For the cloud server, it reads data from dynamoDB, and will separate the data pack into two parts, one is sensor data, another one is about customer behaviors (if there is any), including the frequency that customers come, the time every customer stay, whether and how hard does every customer fiddle with the food, and the amount of food that every customer takes. Then the data about customers is feed into a machine learning model to predict whether customer like this food.
Finally, after all the data is generated, the server will push information of food remaining, customers’ attitude to dynamoDB, and alert manager if the food is not enough or need to be replaced (related to the time since last refill and temperature);
The web server is hosted on amazon S3 bucket using static hosting method. The javascript will use chart API combined with above stored dynamoDB data to create chart for food consumption speed and food remain stock. There is also a js loop interval function to dynamically update the result to achieve real time data displaying. We further decide corrupted data will be dropped.
On the raspberry pi, we use a camera to track customers’ faces, and record the time they come and stay. The raspberry also process and upload the preliminary data;
For the cloud server, it reads data from dynamoDB, and will separate the data pack into two parts, one is sensor data, another one is about customer behaviors (if there is any), including the frequency that customers come, the time every customer stay, whether and how hard does every customer fiddle with the food, and the amount of food that every customer takes. Then the data about customers is feed into a machine learning model to predict whether customer like this food.
Finally, after all the data is generated, the server will push information of food remaining, customers’ attitude to dynamoDB, and alert manager if the food is not enough or need to be replaced (related to the time since last refill and temperature);
The web server is hosted on amazon S3 bucket using static hosting method. The javascript will use chart API combined with above stored dynamoDB data to create chart for food consumption speed and food remain stock. There is also a js loop interval function to dynamically update the result to achieve real time data displaying. We further decide corrupted data will be dropped.
Learn More?