A9: Final Project — Bar Minder

Intro

The final project in HCDE 498: Introduction to Prototyping was a 2-week open-ended exercise in applying a few of the prototyping techniques we had learned throughout the quarter. My classmate Jess Landquist and I decided to partner up to create Bar Minder (working name) as a visual cue system that allows bartenders to better identify customers who need service, particularly in a dark and busy bar environment.


Concept

The concept for the Bar Minder system came from a brainstorming session between myself and Ryan Wills of Dallas, Texas. Initially we talked about a lit glassware system for entertainment purposes that would produce a light show of sorts as patrons lifted and set down their drinks along the length of a bar. Jess and I stepped back from this idea and talked about how a similar concept could be adapted to solve a problem in a bar or club instead of simply using the system as a novelty. We ultimately arrived at the idea of a sensor system that would detect when a glass became empty and visually alert a server or bartender to a customer in need of service more obviously than simply an empty glass that may be difficult to see in a dark environment. A secondary feature would be to help bartenders identify customers that had been waiting longer than others and give them the ability to serve those customers sooner.


Goals

The prototype was intended to allow us to evaluate both the desirability of the system by patrons and to evaluate the feasibility of the system. We chose to scope the desirability evaluation to the patrons first so we could understand their feelings toward the light, such as whether they found it too obtrusive or how they might see it affecting their drinking habits. We also wanted to evaluated the feasibility of creating the system in terms of functionality and potential sensor methods.


Implementation

For proof of concept purposes we chose to create a tabletop unit with two opposing place settings, each equipped with an infrared sensor and a raised guide for glass placement. Behind each guide is a track that allows the infrared sensor to be moved to a set distance away from the glass for fine-tuning the sensor placement. LEDs protrude through a hole below each glass and are connected to Arduinos hidden beneath the table top.

For work allocation, the prototype was split up into two components, the physical design and creation of the table top unit and the programming and implementation of the Arduino/LED system.

Tabletop System Design
I designed and built the tabletop unit, sensor slides, and sensor housings. After sketching out some concepts, I designed the full system in SolidWorks. The dielines of the tabletop components were then extracted to be used for laser cutting the 1/4" white acrylic sheets chosen for their structural rigidity and opacity (and clean aesthetic, of course). The joints in the table were designed to be tight enough to allow us to forgo using glue and be able to collapse and flatpack the table for transport if necessary.

The sensor housings were 3D printed in PLA and used to hide and protect the sensors, which were mounted on two layers of acrylic and bolted through the tabletop along the guide tracks to get them up at a height that would clear the bottom of a glass.

We also made special glassware for the system. While the system works with any clear glass where the sensor is located above the solid base of the glass, we chose to etch mason jars above the sensor line in order to diffuse light and allow the glow to be seen in a more well-lit environment. This was mostly due to the system being exhibited during the day in a space where the lights would be at full levels.

Arduino Coding and Setup
Jess wrote a program in C/C++ using the Arduino integrated development environment (IDE). The program allowed the infrared sensor to detect the absence of liquid in a glass and trigger an LED to illuminate the glass. The program also sequentially stepped the LED through three phases: 1) a solid white light, 2) a pulsing white light, and 3) a pulsing red light dependent upon a function of time. We discussed at length the best light sequences to implement and chose the soft, pulsating pattern rather than a flashing light to minimize the obtrusiveness of the system in a dark environment. Stepping through solid, to pulsing white, to pulsing red was our solution to allowing bartenders to identify the length of time a customer has been waiting in relation to others in the bar or restaurant.


Evaluation

We evaluated the prototype based on desirability by the users we tested it on and the feasibility of building such a system.

Desirability
It should be mentioned that we brought this system to bar for testing and for shooting a demo video. This enabled us to view the system in a real setting under lighting conditions that could be expected in most bars. Although we had a couple volunteer users that we brought with us, we also received comments from some other customers of the bar who were curious about the prototype. In all cases, we received very positive feedback as well as some interesting comments about the system.We were successfully able to determine the desirability of our product based on positive feedback we received through casual conversations with various people, including bar patrons. Both the users and other patrons of the bar expressed a desire to see the system integrated into a bar having experienced scenarios where service had taken much longer than desired. A few also noted the potential of the product as a ‘drinking game’ wherein people race to get the light to turn by finishing their drink first. Obviously, this poses some unintended consequences of the product, but it is an interesting scenario to explore. As expected, users seemed to express a preference for a softer pulsating light rather than a strobe effect, but discussed the option of customizable light patterns based on the nature of the environment (e.g. a club versus a subdued pub). Overall, our evaluation indicated that the system would in fact be desirable, at least by customers. Further research needs to be done to understand how bartenders and servers may feel about the product.

Feasibility
During our ideation phase, we discussed several options for sensing the liquid level in a glass. Ultimately, we arrived at infrared sensors because we had access to them and our timeline was very short, and also because they would allow for some flexibility in placement vertically along the glass. We were pleasantly surprised when they worked better than expected at detecting when a glass was empty. The ability to position them closer or further from the glass using the tracks cut in the board proved crucial for getting the sensors to work as intended and adjusted on the fly if the sensors began malfunctioning. The tabletop system was relatively unobtrusive to the users and they were able to easily place the glasses in the necessary spot to trigger the light. While we would prefer to iterate on the system to integrate the sensors and lights into the glass itself and remove the necessity of a specialized tabletop, the proof of concept indicated that such a system is entirely feasible.


Analysis

The Bar Minder prototype largely worked as expected. It differentiated between the absence and presence of liquid, as long as it had at least some color tint to it; unfortunately, water did not work with the infrared system. The track feature we implemented effectively allowed for fine-tuning of the sensors since each had unique sensitivities despite adjustment of the on-board potentiometers.

The usability testing and general exhibition of the prototype yielded excitement and enthusiasm, as well as lots of discussion of potential features and improvements. We find this feedback and the discussions it generated an indication a product with at least some potential. Below are some of the notes and improvements revealed during testing and discussions.

  • Many people asked about how it worked with water. As mentioned, the infrared sensor does not detect water or other clear liquids. Future iterations may explore a weight-based system that would not be color-dependent.
  • The tabletop setup is too bulky and limiting for widespread adoption. Future iterations will explore a system that incorporates all sensors and lights into the glass itself so that the system could be used with existing tabletops. This would also necessitate a different sensor system, since we found that the infrared sensor did not work when placed directly against a glass.
  • The promotion of binge drinking was a concern of at least one person who inquired about the prototype. To address this issue, we have discussed the idea of using light color to indicate when the system has gone through a certain number of cycles and alert a bartender that a customer has had a potentially dangerous amount of drinks. A counterpoint by one of our test users was that this may be perceived as a form of policing and would render the system undesirable by customers and could result in a public shaming scenario.
  • Poor sight lines in some bars and restaurants may make a quick scan of all customers difficult. A mobile application or a program integrated into the establishment’s customer management system could make it possible for a bartender to identify tables that need service even when they are unable to see them directly and display the amount of time the customer has been waiting for service based on when the sensor was triggered.
  • Customizable light patterns and colors have been discussed as an option that would allow establishments to the alter the mood based on the environment they desire to create. For example, a dance club may want a more vibrant light display as opposed to a beer hall or wine bar. colors based on the mood and environment in a given establishment. We envision this being a function of a mobile app that would allow the bartender to customize the light pattern of all the glassware simultaneously.
  • Since bars have various levels of service times and patrons have differing needs, allowing patrons to select the level at which the light is triggered has been discussed for future glass-integrated versions of the system. A slider along the side of the glass would allow the user to set the point at which they want service and could be altered during the duration of their visit based on changing needs and preferences.

Below is a demo video of the prototype shot in Hotel Hotel located in the Fremont neighborhood:

Acknowledgements

We would like to thank our instructor, Andy Davidson, and his assistant, Lukas Eiermann; our friends and classmates who encouraged us and gave us valuable critique along the way; Hotel Hotel in Fremont for allowing us to film our demo video in their bar (go there, get pizza); and Mel Landquist, Helen Nesbitt, and Justin Jamison for their patience and support.

Thank you to Jess Landquist for taking charge of the programming, indulging my pickiness, and being an overall great partner to work with.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.