Poop Buster - A Vision Thing Idea

24 Sep 2019

Enter Your Project for a chance to win an Oscilloscope Grand Prize Package for the Most Creative Vision Thing Project!

Back to The Project14 homepage

Project14 Home

Monthly Themes

Monthly Theme Poll

Poop Buster

Poop Buster is a robotic poop cleaner for my children pet guinea pigs (there are 2 of them)

although they roam the garden during day time , they go back to cage with bedding at night

Problem

- guinea pigs poop needs to be manually clean by hand. We selectively pick the poop out and discard them at my garden plants and tree as fertilizer, leaving the bedding material intact

Note: guinea pigs are fully vegetarian and their poop actually is kinda not very smelly, not too hard to clean, can be used to fertilize my garden, but it still needs to be cleaned

well, we can in fact change the entire bedding and not manually picking up the poop but that will be kinda wasteful for the bedding material

eventually, the bedding material needs changing but by using poop buster to clean it we can prolong the lifespan of the bedding material and making it clean longer

end of lifespan of the guinea pig bedding (besides poop, there is also urine, so typically when it is too long duration, eventually we throw it at the garden as fertilizer)

The solution

The solution is to use a X-Y plotter to selective pick up the poop (the mechanism, either it is a suction, a gripper or others, needs to manually built).

i put the plotter (with only one stepper motor attached) on the guinea pigs cage to illustrate the idea. Coincidentally, the size is very suitable

The role of the BeagleBone AI is to perform poop reconigtion with is powerful DSP and EVE cores (look in the reference section for its spec). with its dual cortex-M4 cores it should be able to also be used to control the stepper motor pair on my plotter and the 'poop gripper' mechanism

a quick check on element14 community blog, like this one by Jan Cumps, shows that it is possible to do stepper motor with beaglebone black.

with micro-HDMI output i can also connect HDMI monitor to the beaglebone AI to observe its AI poop recognition in action

The Implementation Process

technically speaking the poop is extremely different from the bedding material so a classical image processing method may be sufficient for the task. However, AI method like machine learning or deep learning is general purpose image recognition method that we can train the device to seek and destroy the target. A bonus step that can be achieved with AI is that the engine can automatically derive the stage of health of the pets by looking at the composition of the poop (although i won't try to do this step for this contest)

References

[1] About BeagleBone AI --> copied from https://beagleboard.org/ai

Built on the proven BeagleBoard.org open source Linux approach, BeagleBone AI fills the gap between small SBCs and more powerful industrial computers. Based on the Texas Instruments AM5729, developers have access to the powerful SoC with the ease of BeagleBone Black header and mechanical compatibility. BeagleBone AI makes it easy to explore how artificial intelligence (AI) can be used in everyday life via the TI C66x digital-signal-processor (DSP) cores and embedded-vision-engine (EVE) cores supported through an optimized TIDL machine learning OpenCL API with pre-installed tools. Focused on everyday automation in industrial, commercial and home applications.

Processor: Texas Instruments Sitara AM5729

Dual Arm Cortex-A15 microprocessor subsystem
2 C66x floating-point VLIW DSPs
2.5MB of on-chip L3 RAM
2x dual Arm Cortex-M4 co-processors
4x Embedded Vision Engines (EVEs)
2x dual-core Programmable Real-Time Unit and Industrial Communication SubSystem (PRU-ICSS)
2D-graphics accelerator (BB2D) subsystem
Dual-core PowerVR SGX544 3D GPU
IVA-HD subsystem (4K @ 15fps encode and decode support for H.264, 1080p60 for others)
more ...