TrashTalk

Inspiration

Over 500,000 Americans suffer from some form of motor disability. Consequently, seemingly simple tasks that many of us take for granted become difficult to almost impossible for others to perform. We drew inspiration on this jarring fact and decided to make at least on of these tasks easier: throwing out our trash. While easy for some of us, throwing out the trash requires the proper function of our hands and leg which incited our passion to simplify this process for those incapable of expending two arms to open a trash can and throw out their piece of trash. Moreover, the solution of maintaining an open trash can would most likely lead to insects or bad odors. This is what led us to TrashTalk, the voice-operated trash can.

What it does

TrashTalk is our custom solution to making an accessible trash can by using our voice. By simply stating "Open trash can", our device will open the trash so we can easily throw out our trash without the need of opening our trash can with our hands or a foot pedal. This empowers everyone to be able to throw away their trash without the need for any help or enables the basketball fans to play paper toss without leaving their desk! Finally, we can say "Close trash can" to finally close the trash can once we are done.

How we built it

TrashTalk interfaces a variety of services, languages, and hardware devices in order to function. These mainly include Assembly AI, a Python Flask application, and an Arduino Uno R3.

Assembly AI

We utilized the Assembly AI's real-time API speech-to-text service in order to enable live transcriptions so TrashTalk is always ready to open and close your trash can. We implemented a frontend application using HTML and JavaScript to establish a web socket with Assembly AI's real-time API. Once speech transcriptions are received by Assembly AI via this web socket, the speech is parsed for the appropriate keyword phrases and then a signal is sent to the Arduino to perform the respective action.

Python Flask Application

The Python Flask application serves as the middleman between the frontend speech-to-text service and the Arduino Uno. When the frontend application receives a transcription and parses the text, a signal is sent to the Python app that ultimately forwards this signal to the Arduino using a serial communication protocol (UART).

Arduino Uno R3

We programmed a stepper motor to open and close the trash can through an Arduino Uno R3. Once a signal is received by our Python script, the Arduino spins the stepper motor in the appropriate direction to either open or close the trash can.

Challenges we ran into

Throughout the weekend, we encountered various obstacles that we needed to overcome. Firstly, we wanted to make our product as self-sufficient as possible; however, due to limitations imposed by the hardware that we had available, we were forced to adapt by connecting our Arduino directly to a computer so we would have the ability to call APIs and use a microphone. We also encountered difficulties in constructing our apparatus due to our limited experiences with hardware.

Overall, we faced the challenge of incorporating unfamiliar services into our project, most notably Assembly AI where none of us had previous experience; however, we were able to combine our strengths in order to ultimately overcome these barriers.

Accomplishments that we're proud of

Given that none of us have had hardware experience, we are most proud of our abilities in coming together in order to create something that extends well beyond our comfort levels. Moreover, we are proud of the fact that we were able to manifest something that we envisioned at the beginning of the weekend into reality.

What we learned

In addition to the new services, languages, and products that we used throughout this hackathon, we learned most about how valuable collaboration can be in creating an ambitious and unique product within a short period of time.

What's next for TrashTalk

There are many different avenues for which TrashTalk can expand in the future. Firstly, we could minimize the hardware overhead in our project by using Bluetooth or WiFi modules which would provide a much more portable solution than what we were able to create during this hackathon. We could also utilize Natural Language Processing libraries in order to make interactions with TrashTalk much more seamless and conversational. Ultimately, the ideal goal for TrashTalk would be for it to permeate to everyday smart assistants and other home appliances that make it hard for those with disabilities to operate.

Team Leader: Sohil Kollipara Discord: Sohil#3145