Week 3

Week 3 Notes:

  • Tested the colour sensors using the 3D printed piece to ensure that the sensor would detect the colour of the skittles in the right environment.
  • Managed to get most of the individual 3D parts together to build the 3D model for the CANDI project
  • Tested the emotion detection mechanism ensuring that it reaches 90% accuracy.
  • Printed out the 3D pieces for the mechanical aspect of the dispensing.
  • Embedded AI in the emotion detection mechanism to increase accuracy.

Challenges faced in week 3:
  • Few pieces for the mechanical dispensing had errors in their dimensions so they had to be printed again.
  • Introducing the emotion detection model to a new environment where the lightning is different took sometime to re-calibrate . The motors were not working properly. However, after spotting out few errors in the code they work perfectly.
  • Try to put the different components into the 3D model was extremely challenging . In conclusion we decided to put all breadboards at the bottom of the machine while both Raspberry pi and Arduino  at the top as both are lighter.
  • The RGB sensor had an issue with detecting the skittles after connecting the 3D piece to the motor so we decided to use a lighter material because the weight could be affecting the motor movement.
Week 4 Objectives: 
  • Add a dock station for the emotion detection where the image would be inserted and will always be in the position
  • Add tubes to the 3D-model that would house the wire so they wouldn't get tangled with other components.
  • Insert the start button for the emotion detection before every image captured.
  • Final testing for the sorting , emotion detection  and dispensing mechanism.
  • Assemble the entire 3D model and add all the parts together.
  • Testing of the final prototype.

Comments

Post a Comment

Popular posts from this blog

Week 1

Image
Background and inspiration  Our project  CANDI (Candy Allocation & Notation Dispensing interface),  aims to sort skittles into containers by using a colour sensor that detects RGB values of the candy and then  dispenses coloured candy based on the facial expression of the person in front of the camera. The  camera detects the facial expression of the person in front of the machine and dispenses coloured candy based on the emotion, i.e., angry = red. We first decided to integrate the Raspberry pi and the Arduino to work coherently, but after some time its been decided that it would be better if each board is independent. The Raspberry pi would be in charge of the emotion detection and move the motor based on the emotion that is most dominant, whereas the Arduino is responsible for the sorting of the skittles. Week 1 objectives: Finalize the 3D design model. Write code for Candy sorting. Write code for Emotion detection. Write code for Dispensing. Test the ind...
Read more

Week 4

Image
Week 4 Notes: Put the entire 3D model together. Tested out the dispensing mechanism with different emotions and added a button to start the emotion detection. Tested out the combination between the first phase and the second phase of the project working together.   Challenges faced in Week 4: We had issues putting the colour sensing and sorting mechanism into the 3D model, we had to recalibrate the movement of the servo that moves the skittle correctly. We also had trouble aligning the funnel for the skittle sorter and had to find the new dimensions for the location of the funnel to make the top of the CANDI machine. Placing the components into the machine took more time than we thought. Due to the number of wires required by the machine, we had to double-check that everything was connected in the correct place.  Conclusion of the Project The machine was successfully able to sort skittles by their colour and dispense them according to different human emotions. We could have wo...
Read more

Week 2

Image
Week 2 Notes: Considering the issues with the colour detector faced in week 1, we found a new way to calibrate and find the correct colours for each skittle, by taking multiple samples of RGB colours of each skittle. However, the detection would work better if it was in a fixed position instead of being handheld.  The SD card for the raspberry pi was flashed and setup so that the code for the face recognition could be tested and implemented using the raspberry pi camera module. Started 3D printing individual components of the machine. Week 2 Objectives:  Test out the colour sensor in different environments to calibrate and find the most accurate thresholds to ensure accuracy in the sorting.  Run the facial recognition tests to determine the efficiency of the raspberry pi camera module and if it detects the emotions correctly. Attempt to print out the 3D components. Challenges faced in week 2: The LED keeps flickering whenever the servos move due to the voltage drop, which...
Read more