The problem of plastic waste floating in rivers and lakes is increasing. This leads to water pollution or non-biodegradable plastic waste that will clog and cause loss of aesthetics. Therefore, our team's vision is to design and manufacture a robot that automatically moves along a route on water. It will be compact so it can be wriggled and collected.
Garbage Robot has a name that is Clync. It is a system for cleaning trash on water (rivers, lakes), operating in an autonomous manner that is set up based on where and when the user wants to clean.
In this project, we develop many different segments with different languages for their specific purposes. Below are the languages used in this project.
We use Flutter Framework for cross-platform application development for web/android/ios which helps in optimization and better UI development and stability. Express.js simplifies handling HTTP requests and routing management. It is used with NodeJS.
Our team uses the STM32F303RET6 microcontroller to process input data, merge values and filter noise of modules. Additionally, the use of Raspberry Pi 4 for object (garbage) recognition processing using YOLOv5.
This device operates on water (lakes, rivers) so cannot use Bluetooth, Wifi, Zigbee and other short-range wireless communication technologies. Those types consume more energy so they are not suitable for the operating range of this device.
LoRaWAN is a wireless communication technology used for wide-range IoT (Internet of Things) networks (LPWAN - Low-Power Wide-Area Network). LoRaWAN is one of the technologies that provides IoT device connectivity with low power consumption and can serve long communication distances.
LoRaWAN technology uses the LoRa (Long Range) wireless communication protocol to connect IoT devices to network infrastructure through communication gateways.
We decided to choose LoRa as the primary wireless communication from End device to Gateway.
Overview of System Architecture
1. Application Layer
- Showing data to Web/App to track position of Robot moving in the river.
2. Data Processing Layer
- Taking data from The Thing Network Server and sending to NodeJS server with MongoDB (Local Storage) through MQTT
3. Transport Layer
- Transfer data to Gateway Lora through LoRa
- Data flow like this [End Device] -> [Gateway] -> [The Thing Network] -> [NodeJS Server]
4. Perception Layer
| Type | LoRa RFM95W | GPS NEO 6M | MPU6050 | HMC5883L | BTS7996(Actuator) | Raspberry Pi 4 |
|---|---|---|---|---|---|---|
| Protocol | SPI | UART | I2C | I2C | GPIO(PWM) | SPI |
| Voltage | 1.3-3.3 | 3.3-5.0 | 3.3-5.0 | 3.3-5.0 | 3.3-5.0 | Empty |
Use Interrupts and Asynchronous - Develop an algorithm that follows point coordinates in a point list
1. Start new project
npm init
2. Install dependencies
npm install body-parser, express, mqtt, mongoose, socket.io
mqtt - MQTT protocol for trasfering and receiving data from Network Server (The Thing Network - MQTT Broker)
socket.io - Using for Real-time application between Server and Application Web/App
mongoose - Connecting with Database MongoDB Compass in Desktop
3. Connect Local Server to The Thing Network through MQTT
- First, declare mqtt variables for mqtt dependencies
var mqtt = require('mqtt')
- Second, add the appropriate required configurations
const TTN_BROKER = 'mqtts://nam1.cloud.thethings.industries:8883'
const TTN_USERNAME = 'autonomousrobot@autonomousrobot';
const TTN_PASSWORD = 'NNSXS.ITEXEECGUFE3TVGKRXNPTRHVP7FMRDXQT3VFDFA.72UEPZWMCLPJOIVBMMKUGJ7UEIZPOH6FWAGI2EWUDUNEAZQUOI2Q';
const APP_ID ='autonomousrobot'
const DEV_ID ='fuckyoudat'
- Option contain USERNAME and PASSWORD requirement
var options = {
username: TTN_USERNAME,
password: TTN_PASSWORD,
keepalive: 60,
reconnectPeriod: 1000,
protocolId: 'MQIsdp',
protocolVersion: 3,
clean: true,
encoding: 'utf8'
}
- Connect to MQTT Broker ( The Thing Network )
var client = mqtt.connect(TTN_BROKER,options);
- Listen event "connect".
client.on('connect',function(){
console.log('-> The Thing Network MQTT Broker Connected <-');
/* .......
.......
TODO: IN HERE
*/
})
- Continue 'TODO: IN HERE' - Subscribe to receive uplink message from TTN.
// Subscribe to receive uplink messages
const uplinkTopic =`v3/${TTN_USERNAME}/devices/${DEV_ID}/up`
client.subscribe(uplinkTopic, (err) => {
if (err) {
console.error('Error subscribing to TTN:', err);
}
});
- Continue 'TODO: IN HERE' - Subscribe downlink to push from Server to TTN.
// Định nghĩa topic downlink
const downlinkTopic = `v3/${TTN_USERNAME}/devices/${DEV_ID}/down/push`;
// Tạo payload
const message = {
"downlinks": [{
"f_port": 3, // Chọn FPort thích hợp
"frm_payload":"AQ==" ,
"confirmed": true, // Chọn true nếu muốn xác nhận downlink
"priority": "HIGHEST" // Mức độ ưu tiên của downlink
}]
};
- Handled event "message" when catching.
client.on('message',function(topic,message){
// Handle the incoming message
const payload = JSON.parse(message.toString());
console.log("Data from TTN: ",payload);
GPSService.gpsFromTTN(payload)
})
Developing application using Flutter (Cross-platform for Web/App) and intergrating Map API of MapBox Service. I guide for embedding url link to Map package Flutter Map.
URL = "https://api.mapbox.com/styles/v1/treehot68/{mapStyleID}/tiles/256/{z}/{x}/{y}@2x?access_token={accessToken}"
mapStyleID = Creating Map Project in Map Box, it will give you a link contain 2 important parameters that are Map Style ID and Access Token
accessToken = Same above
Before showing the important information into the Website, The Thing Network platform must receive package data from the Gateway. After that, it is transfered to NodeJS Server to handle and choose the pivatol data through MQTT protocol.
The picture below is Dashboard Website for monitoring the position and the information of end device.
Dashboard in Website