QRS Robotic Violin

Summer 2018 and Summer 2019

Virtuoso Violin paired with player piano

QRS and the Task:

QRS is a player piano/automated instrument company based in Pennsylvania. Their "Virtuoso Violin" is meant to be an accompaniment to their player pianos, so that the two instruments can play duets together. However, the Virtuoso did not sound like a real violin, and due to its complicated design, the violin would break frequently.

The work to improve the Virtuoso Violin was initiated by the CEO of QRS, Tom Dolan, and Tufts Professor of Music Engineering, Paul Lehrman. Professor Lehrman assembled a team of undergraduate students to take on this effort. Our team began work in the Summer of 2018, beginning with taking apart the Virtuoso Violin and working to rebuild it as a much more robust and streamlined system. After a successful summer, work on the project has continued into the 2018-2019 school year.

Original Design:

The sound of the violin was actually not produced at all by the bow mechanism. QRS Music has a patented transducer they embedded in the bridge of the violin. When powered, the transducer would emit the "violin-like" sound, amplified by the body of the violin. The bow mechanism was more for show, to give the illusion that the violin was actually being played. One of the constraints we were given was that our new design still had to make use of the transducer, so we focused on creating bow motion similar to how a human would play the violin.

The original design of this mechanism was complex and unreliable. As the "lead roboticist" on the project, I identified three things for us to focus on that summer:

Bow Drive Complexity: The original bow drive with fishing line and pulleys was too complex and easily tangled or broken. We needed to scrap this design and create something simpler from the ground up.
Lack of MIDI Compatibility and Programming Errors: It was also controlled by MIDI commands, which are meant for controlling the sound output of electronic instruments. As is, it could not properly communicate with the motors that drove the pulleys. We needed to find a way to convert the MIDI commands it was receiving from the computer to the analog input that motors can work with.
Reliance on Expensive, Custom Parts: Upon first inspection, we immediately realized that this had been overdesigned. The product did not need a huge brushless motor, or the numerous custom boards at work. We identified replacing the original hardware with cheaper and more robust off the shelf options, and replacing the program with open-source and at the ready software, as a priority.

Complex Bow Drive Mechanism

Lack of MIDI Compatibility

Reliance on Expensive Custom Parts

Summer 2018 Design Work:

I completed the following tasks:

Simpler, Robust Bow "Cradle": The redesign incorporates a 3D printed cradle that houses both the bow and a simple DC motor. A toothed rack runs along the length of the bow. This rack is driven by a gear connected to a motor by a timing belt. This creates linear motion of the bow. Bow angle is controlled by a servo motor connected to the bow with a thin shaft. This assembly was much simpler than the previous pulley system, and much easier to repair if something went wrong.
SparkFun MIDI Shield and Proportional-Integral Control Software: The SparkFun MIDI shield mounts directly onto an Arduino board and allows the Arduino to recognize MIDI. Through this shield, we can receive MIDI commands, which in turn will have the Arduino send analog commands to the motors. In addition, an entirely new Proportional-Integral control was implemented in the Arduino code to better control the bow drive.
Simplified Motor and Circuit Requirements: The robotic component of the current violin is run entirely off of the Arduino Uno board and the circuit pictured above, as well as a smaller, simple DC motor. In production, we would need to get rid of the breadboard and create original PCBAs, but this shows that it would be a huge reduction compared to the original design.

An electrical engineer on the project also completed work improving the sound output from QRS' transducer.

The team presented our work to the CEO of QRS, Thomas Dolan, who was thrilled with our work, and funded another summer of work on it in 2019.

Simpler, Robust Bow "Cradle"

SparkFun MIDI Shield

Simplified Electrical Requirements

Summer 2019 Design Work

I continued work on the project during the Summer of 2019. Many design changes were made to further improve the movement of the bow.

After taking graduate level coursework in control theory, I further refined the control software so the bow reaches its final speed at a faster rate, and also aides in keeping the bow speed and motion stable.

The cradle design was simplified. Rather than a tooth rack, the motor is connected to a 3D printed shaft that acts as a spool for a line of guitar string. This is much easier to set up, and provides a more fluid motion of the bow similar to how a violinist would move it.

Two magnets were inserted inside the bow shaft itself (which was custom made using carbon fiber), and two hall effect sensors were added inside the cradle. When one of the sensors triggers, it tells the bow to switch direction. This acts as a safeguard against the bow going off the track.

Reflections:

Looking back, there are two main areas I would focus on if I were to revisit this work:

See if there is a way to move the bow drive mechanism completely inside the box that holds up the violin, to better create the desired "illusion"
Much of this work was proof of concept. I would love to go back with the lens of someone who has worked on consumer products and worked with manufacturers and see how we can create a high fidelity version of this violin that could be seen on store shelves.

Google Sites

Report abuse