Schedule
3/10/2016: Prepared project for final demo. All subsystems completed except for parts of HMM algorithm.
3/8/2016: Completed construction of glove. Finalized MIDI communication.
3/3/2016: Continued work on MIDI communication. Achieved successful MIDI communication.
3/1/2016: Received 3-D model of our glove box. Installed circuit components into the box. Finished soldering LEDs to the leads of the circuit. Continued work on MIDI without any progress.
2/24/2016: Still receiving errors due to faulty MIDI recognition. Analyzed raw data sent from the Raspberry Pi. Have received complete MIDI signals although they were not the signals that were supposed to be transmitted.
2/17/2016: Began the training process for the HMM using the training data collected last week. Continued debugging the HMM algorithm.
2/10/2016: Attempted to send MIDI but have experienced faulty MIDI recognition. Continued to debug HMM and have corrected several issues with the simulation. Acquired test data for the HMM.
2/3/2016: A serial data connection has been established. Data can now be sent from the Raspberry Pi to the DJ's computer. We will now be working on sending the appropriate MIDI signals to the DJ software. Several more components were acquired and installed for the glove. Object data has been extracted from the Pixy system for testing in the MATLAB HMM simulation. Once several gestures have been successfully tested in the MATLAB simulation, we can begin transferring the code to C.
1/27/2016: We analyzed MIDI signals transmitted from a Numark MIXTRACK DJ board with the MIDI monitor application. We also designed a layout for the perf board and began soldering. Measurements were taken for the circuit enclosure and a design is currently being created. Work has been done on extracting the relevant object data from the Pixy camera onto the Raspberry Pi.
12/8/2015: We have finalized the circuit for the glove. Glove design and building has begun and will continue next semester. SPI communication has been established between the Pixy and the Raspberry Pi and we are able to extract data about the objects detected from the Pixy onto the Raspberry Pi.
12/1/2015: Work has been begun on SPI communication between the Raspberry Pi and the Pixy. The code that updated the parameters of the HMM has been completed. Work on the LED circuit has been postponed because we are waiting for parts to arrive. Work on the Mixxx communication has also been postponed until we acquire a MIDI controller.
11/17/2015: We have begun conceptualizing the glove design. There has also been work done on improving LED tracking of the glove. Although there is still testing to be done, we finalized the design of the circuit. The Viterbi algorithm was coded and tested successfully. Research has been completed for communication between Pixy and the Raspberry Pi.
11/10/2015: Designing the PNP transistor circuit did not meet the goals of the LED circuit. We went back to an earlier design to see how Pixy object detection worked, and the results suggested that the design is appropriate. The object detection is initially successful. We also began drafting XML and JavaScript code for Mixxx software application as well as continued experimentation with Pixy LED detection. More work was done on the Viterbi algorithm as part of the process towards a complete HMM simulation.
11/3/2015: Began drafting sample Javascript and XML files for the Mixxx DJ software plugin. Completed work on the Viterbi algorithm for the MATLAB simulation. Forward and backward algorithms were tested to ensure that they worked correctly. Work was begun on integrating all HMM algorithms into a cohesive simulation. We have begun work on designing a PNP transistor circuit because there is a problem with current draw during both the saturation and off modes of the NPN transistor. The PNP transistor will limit the current draw that we don't desire while keeping the current draw that will give us the bright LEDs. We designed the blue diode transistor circuit to match the 20 mA current specification. The next step for the LED circuit is to design for the green and red transistor circuits and debug the mode select portion.
10/28/15: The forward and backward algorithms have been completed in the MATLAB simulation. We are currently working on the Viterbi algorithm as well as initializing the HMM. The LED circuit and current amplification circuitry is being constructed and designed right now.
