Embedded Systems

Welcome! You'll find cool embedded systems projects here!

Learn More

About the Project

This project was focused initially to serve as a prototype to aid the physically challenged through a voice recognition model. The tiny RC car mimics/represents a system that could provide movement for the disabled through the use of simple commands such as start/stop vehicle. The control of the direction could be implemented using advanced ML features such as eye tracking, object tracing or through joysticks that could be actuated from other parts of human body such as jaw movement. The tiny RC car gives a minimalistic feel as to how this could be implemented so that it could be expanded further in the future. Feel free to get in touch to learn more or to work together! ;)

Go to Projects
...
...

RC Car with Obstacle Avoidance and Voice Recognition

The tiny car is equipped with ultrasonic sensors that override user input from the remote controller to prevent any collisions with hurdles in the way. A voice recoginition model has been implemented to 'Start'/'Stop' the car upon user input.

...

Project Instructions for re-creation

To compile the code, please follow the following steps: 1. Setting up the remote: To set up the remote, please compile the code inside the transmitting folder inside the repository. Use adafruit qtpy for the transmitter. ALso, use another pico4ml board inside the remote control for the voice recognition system. This contains the usage of the pio module. Please use cmake to compile this. 2. Setting up the receiver: We are using two modules here: one rp pico and one pico4ml board. The rp pico receives the bluetooth information and sends it to the pico4ml, which is used to control the wheel speed of the car. Another python file is provided in this folder, that contains the implementation of the motor driving system. This must be compiled using circuitpython. For information regarding reflections on design, features found satisfying and PIO explanation, visit https://koushik-sss.github.io/team-roadrash_project/

...

Significant Troubleshooting

One significant challenge in our project was to make the voice recognition model to be immune to ambient noise. This was done by using 2 categories of sounds to be classified: noise and stop. The noise samples are taken from the real life noise samples which were collected using the microphone of the Pico4ml board. Another challenge we faced was the sampling rate of the microphone. We experimented with different combinations and finally arrived at 16kHz sampling rate for the microphone. Another significant issue was the calibration of the joystick. The joystick provides output from 0 to 1023 with a value of 512 for the rest position. However, the joystick was not calibrated and hence we had to follow the following logic to calibrate it, since the rest position was 756, instead of 512. For information regarding reflections on design, features found satisfying and PIO explanation, visit https://koushik-sss.github.io/team-roadrash_project/

Subscribe to receive updates!

Address

6N 42nd St, USA

Email

surajsbm@seas.upenn.edu koushik@seas.upenn.edu

Phone

(929) 613-7802