Ramudroid - Trash picking robot

Ramudroid prototype

alt Ramudroid

Ramudroid is a bot to clean roads and outdoor environments. For brains, there's a Raspberry Pi onboard and communication is via wireless ( Wifi + BLE) connectivity. The camera module provides capture for WebRTC streaming and image sensing of the environment on the remote end.

Picking up trash ( like used masks, gloves, napkins or even regular objects like used paper cups) are hazardous to the health of sanitization workers . A robot meant to pick up litter essentially keeps the workers away from getting in contact with potentially contaminated trash.

Robot's Hardware

alt Ramudroid

List of components

Processing and MicroController

Raspberry Pi 3B+/4 as Central Processing Unit

Web services to receive remote navigation command
Uv4l streaming server for webrtc
OpenCV remote object identification

Arduino Uno as Microcontroller

receives commands from Rpi on a serial interface
control motors for movement and cleaning

Driving Unit

Pi NoIR Camera V2 Object and obstacle detection uses Sony IMX219 8-megapixel sensor accessed through the MMAL and V4L APIs,

motor driver L298 Motor Driver

power Lipo battery 11.1 V or Solar Panel connection 12 V

Cleaning Unit

brushes motor 3 x 5V DC gear motor

relay 5V single channel relay

power Lipo battery 11.1 V or Solar Panel connection 12 V

IR sensor to detect if garbage collection bin/tray is full

infrared radiation can be found between the visible and microwave regions, wavelengths between 0.75 and 1000µm.

Software

Following are the modular components of the project :

1. m2m-communication communication

Communication between the web client, mobile client, cloud server and robot's core unit is primarily on REST API's. Communication techniques used in the project are as follows are ,

GPIO
UART
I2C

2. Computer Vision, object tracking, real-time video analysis

To effectively limit the usage of power on frontal cleaning brushes, it is crucial to target garbage/litter and active the motors only when a suited target is found. Image analysis based on Haar and SVM classifiers to detect garbage such as bottles, masks, cans.

3. Robot's core control unit

Rpi connects the Motors and their drivers via Arduino. It connects to sensors for input and batteries for power while controlling the operation. and coordination between remote instructions and robot GPIO access library, written in C for the BCM2835, is used in the Raspberry Pi to control pins.