Table 1.

Design specifications and limitations of implemented system model.

Part Name	Details
Robotic Arm	Task	The core functionality of this robotic arm is to pick a single item from a trash bucket and place it on the conveyor belt.
	Specification	It can pick and place trash item of around 2 KG. The arm is capable of shifting 15 items per minute.
	Limitations	Claw of arm is capable of picking items of 12 cm to 20 cm. The weight limitation of arm is 2 KG.
Conveyor Belt	Task	Functionality of moving conveyor belt is to move within a specific speed range defined as 12 complete rotations in a minute.
	Specification	Trash items will be moved after classification to their corresponding buckets. It can carry items of around 2 KG in weight.
	Limitations	Speed of the movement cannot be increased in this step, because this architecture is also configured with the classification process, so the specified time interval is also added to the total time of rotation. Weight limitation cannot be increased in this step, because this will also slow down the belt movement process. Using heavy motors and refined mechanical model can increase the weight limitation.
Camera	Task	Capturing images of the trash items and transferring them to the connected Raspberry Pi module for classification process.
	Specification	It has GPU support that increases the overall image capturing and transferring process.
	Limitations	There is no limitation for this module of the system.
Raspberry Pi	Task	A tiny processor providing control structure for the classification process. With connected local server configured on Jetson Nano, it can transfer required data to the other module of the architecture for the next process.
	Specification	Can run machine learning model for classification process. A portable device that is easy to program and configure.
	Limitations	Low computation power can affect the performance of the local system, not over-line.