Environmental Visual Examiner (EVE)

College of the Canyons Association of Computing Machinery Research Project

View our project abstraction for a summary of how this system should work. To view more of the research completed so far, view our google drive

Design

This configuration is using AWS IoT, DynamoDB, and Lambda services from AWS. A pre-defined eve_user table in DynamoDB holds information about the user's garden plot for querying. An MQTT client is setup through AWS IoT to subscribe and publish from/to the raspberry pi at scheduled intervals using configuration details from sensor_reader.py (certification must be created and linked to the pi using the topic thing01/data). A rule is then used to insert published sensor values into our DynamoDB eve_data table.

From DynamoDB, our water_alg Lambda function is triggered, which requests eto from the CIMIS API, queries the plant factor from our eve_pf table, and calculates whether watering is required. The results are then written back to the same item in our eve_data table. Lasty, an MQTT message is published back to the pi with watering instructions, which are then executed.

DynamoDB

Three tables are used for the execution of this pipeline, their names and schema are listed below, strings are used for most of the attributes to maintain the integrity of the data when storing:

eve_user - pre-defined values, create an item for testing

• user_id: String (PRIMARY KEY)
• location: String (SORT KEY)
• date_created: String
• plant_spec: String
• plant_type: String
• plot_size: String

eve_pf - uses information from UCANR resources

• plant_type: String (PRIMARY KEY)
• plant_factor: Number

eve_data - lists with sensor, user, and alg info are inserted into this table

• date_time: String (PRIMARY KEY)
• data: Dict
  • date_time: String (PRIMARY KEY)
  • temp: Number
  • humid: Number
  • light: Number
  • moist: Number
• user_info: List
  • 0: String (user_id)
  • 1: String (plant_type)
  • 2: String (plant_spec)
  • 3: String (plot_size)
• water_alg: List
  • 0: String (pf)
  • 1: String (eto)
  • 2: String (gal_water_reserve)

Lambda

When configuring the Lambda function, a DynamoDB trigger must be selected to invoke the function when an item is inserted into the eve_data table. A deployment package (zip) must also be created which contains the dependencies for the function installed directly into the project folder. To do this, run pip3 install requests -t <project_folder> The zip must then be uploaded to Lambda and unpacked so that AWS may find the main.py script. The handler for the Lambda function is main.water_alg