This work proposes an autonomous waste separation system in three classes (Organic, Plastic and Paper) as a way to optimize the recycling process, incorporating an artificial intelligence algorithm that allows to recognize and classify the different waste in the mentioned classes by means of a mechanism of pulleys and motors so that these are deposited in the right place, without the need for human intervention, it is also proposed to design as complements to the general mechanism, a chatbot that communicates with the cleaning staff at the time when the containers (bins) are full. For the first part of the proposal, i. e. , the algorithm, a dataset created by the team is used in which photos of a large amount of solid waste are found, and through deep learning techniques a CNN (Convolutional Neural Network) is trained with this dataset where the 3 classes of waste that are handled are specified. For the second stage of the proposal, a mechanism of pulleys with toothed chain is used to transport a capsule (containing the waste) through the general structure, this movement is performed by a DC motor and the stopping points are delimited with reflex infrared sensors, this process begins when the AI (Artificial Intelligence) mentioned in the previous point, communicates with an Arduino and sends it the data of what type of waste has just been deposited, consequently the motor of the mechanism is activated and when the capsule is in the right place, the lower zone of the mechanism opens by means of a servomotor and the waste falls due to gravity, finally the mechanism returns to the initial position. Finally, the chatbot is used as soon as the cleaning staff receives a notification via Telegram. It can be seen that the sum of the deep learning algorithm together with the internal deposit mechanism, allows an optimized waste separation process that culminates in a good recycling process.
This work proposes an autonomous
waste
separation system in three classes (Organic, Plastic and Paper) as a way to optimize the recycling
process
, incorporating an artificial intelligence algorithm that
allows
to recognize and classify the
different
waste
in the mentioned classes by means of a
mechanism
of pulleys and motors
so
that these
are deposited
in the right place, without the need for human intervention, it is
also
proposed to design as complements to the general
mechanism
, a chatbot that communicates with the cleaning staff at the time when the containers (bins) are full. For the
first
part of the proposal,
i. e.
,
the algorithm, a dataset created by the team is
used
in which photos of a large amount of solid
waste
are found
, and through deep learning techniques a CNN (Convolutional Neural Network)
is trained
with this dataset where the 3 classes of
waste
that
are handled
are specified
. For the second stage of the proposal, a
mechanism
of pulleys with toothed chain is
used
to transport a capsule (containing the
waste)
through the general structure, this movement
is performed
by a DC motor and the stopping points
are delimited
with reflex infrared sensors, this
process
begins
when the AI (Artificial Intelligence) mentioned in the previous point, communicates with an Arduino and
sends
it the data of what type of
waste
has
just
been deposited
,
consequently
the motor of the
mechanism
is activated
and when the capsule is in the right place, the lower zone of the
mechanism
opens by means of a servomotor and the
waste
falls due to gravity,
finally
the
mechanism
returns to the initial position.
Finally
, the chatbot is
used
as
soon
as the cleaning staff receives a notification via Telegram. It can be
seen
that the sum of the deep learning algorithm together with the internal deposit
mechanism
,
allows
an optimized
waste
separation
process
that culminates in a
good
recycling
process
.