When Walking Into A Park, Where Do You Sit?
Sensing Urban Life in Place

Spring 2019  
GSAPP Elective, Urban Informatics II, Columbia University
Instructor: Anthony Vanky
Group Work with Xinyu Liu, Kevin Kim



In New York City parks, there are 43 bench seats per acre. However the relationship of benches and where people sit has not been measured and studied thoroughly. In most circumstances, as William Whyte stated, “people tend to sit most where there are places to sit.” Does this hold true in public spaces when people are given choices about location, type, and orientation of benches? Behavioral cues of people sitting on benches can provide many valuable
information not only on understanding people’s patterns in public spaces, but it will also help architects and planners to effectively allocate sitting placements.

In order to understand sitting preferences and patterns of park users, our group collected sample data on people using the benches. On May 4th, we dispatched and installed ultrasonic sensors on to three benches and collected data on people sitting on the benches. With collected data, our group analyzed:

1) the use rate of seats;
2) the use pattern in terms of time, location, and types of seats.

Motivated by these questions, we believe that the best approach to design sitting spaces in a desirable way is to first observe and, in this case, to sense people’s actual behavior and their interaction with these spaces or settings. This is essentially what motivates us to launch this project, though as a prototype using low-cost sensors, to monitor and evaluate the use of sitting spaces in existing parks using a simple and cheap method to achieve a better understanding about people’s sitting behavior.

Technologies

To achieve this local interaction, our main task is to detect human presence on benches. The technological challenge we confront is to select one type of sensor that is stable and reliable, less vulnerable to weather conditions, and not so conspicuous that may impact people’s decision-making. We come up with three alternatives: infrared sensors, vibration sensors, and ultrasonic sound distance sensors.

With a number of pilot experiments, we selected ultrasonic sound distance sensors , as it is less susceptible to the surrounding light, color, and motion. The sensor’s working principle is to transmit a high-frequency signal and measure the distance between the sensor and the object by calculating the time interval between sending and receiving the signal.




To detect the local interaction, (i.e., the human presence on a bench), the sensor will be installed to each of the benches to sense and monitor whether people are taking the seats. These small installations will solicit data, and they are not intended to have any influence on people’s decision-making process. 

Another imprtant part for the local interaction is how to use the data for our analysis. As the ultrasonic sensor
will generate lists of distance from an object, it helps us identify where objects fall on the bench, or where people sit. For instance, if we know the length of the bench, by looking at the fluctuation of value, we can know if a person sits in the middle, or at the edge.

Lastly, we look at how to install the sensor so it could work appropriately including test the height, and different way to orient the sensors.

Design Iterations

Iteration 1-3

Iteration 1: Naked Sensors
Like all the great inventions in the world, we start with naked installations: a board, a sensor, and a few wires. We use tapes and rubber bands to attach them to each other and fix them onto benches just in front of Joe’s Cafe on campus. Data collected were accurate and responsive. Unfortunately, this really looks like a bomb and it will be likely to scare people off!

Iteration 2: Minions
Looking at the ugly, naked sensor, we got inspired by the two cylinders (transmitter and receiver of the sensor) and came up with an idea to “cutify” the installation. We used foam boards with the image of minion “Kevin” and adhere the sensor to its back. In this way we hope that though being more visible, these sensors may be less scary and people will not, probably, avoid them.

Iteration 3: Sensor in Cube
However, there remain some concerns about people’s reaction to them.
Taking all of these into consideration, we made this final version - a small and simple cube with a sensor inside.
We did this design iteration with constantly thinking about how would the interaction be between people and the installation.

Pilot Research / Implementation

The research is going to be conducted in the Starlight Park, located in North Bronx. For the scope of this project, we will primarily focus on selected bench types and will not take other sitting spaces (stairs, ledges, etc) into consideration. After several site observations, we selected three types, six benchs and insalled one ultrasonic sensor on each. Each sensor would provide timestamp and distance of an object located with one second time intervals. This data was captured on May 4th, from 11:00am to 17:00pm,, and acquired totaling 129,600 rows of data in total from six ultrasonic sensors.

To analyze the use pattern, we needed to first understand the measurement of occupancy rate. First of all, since the ultrasonic sensor has a distance limit of two meters, any measurement that went over 2,000 was considered unoccupied.We further restructured the data to a binary data set to simplify measure the occupancy based on time spent on each bench. In addition, the suddden fluctuations of distance while the  numbers are still ranging in occupancy category help us identify if there were more than one person sitting.

Benches in Starlight Park (from left to right):
1: Bench with backrest and table
2: Bench with backrest only
3: Bench with no backrest

Observation

Alongside the deployment, we have observed a few interesting phenomena, which encourages us to think about how the experiment could go further.

First, we installed the sensor along with a public notice, to inform the public that the project is free of private concerns. As people would react differently to the project, they tend to sit if they see someone sat in the place earlier.

Also, the bench which the sensor box was attached to the side has clearly more people sit on it than others. Another interesting observation is how the design of sitting space could be ‘wasted‘ if they are not properly designed. Bench #3 is located near the riverbank, yet the view is largely blocked by the trees and bushes. However, in the back is the soccer field. We notice that people would sit on the rock or lean on the chair rather than using the seats.


Quantitative Analysis

Based on data collected, our group looked at the bench utilization rate (%) and occupation period.

From bench #1 we collected 19,549 rows of data, which accounted for approximately 5 and a half hour worth of data. From bench #2, we collected 10,534 rows, equating a duration of 2 hour 55 minutes. Bench #3 collected 21,515 rows data, which measured 6 hours of data. Bench utilization throughout the three benches demonstrated the around 10% utilization controlling for discrepancy in measuring period. Bench #1 had 8.1% utilization rate, and bench #2 and #3 displayed 11.9% and 9.5% utilization rate respectively.



Continuing from the utilization rate, we applied temporal measures to understand the occupancy patterns throughout the day. Bench #2, which was placed in between the basketball court and soccer filed had the most balanced occupancy of the bench. On the other hand, bench #1 and #3, which were placed toward the edge of the park demonstrated lower occupancy during the early part of the measurement, but the occupancy rate increased as the day passed. From the pattern of occupancy, we were able to understand that benches that are placed in the center of the park was more uniformly utilized.

Although data analysis did not provide much significant difference in utilization pattern due to missing data, it could still be found that the location of benches have impact on the occupancy of the bench.

Moving Forward

This small installation, being easy and simple, can provide the small-size, neighborhood parks with opportunities to understand users’ behavior and evaluate the facilities they offer in a way that generates little cost.

Meanwhile, with a significant amount of data in abroader area, these data can be such a good resource for researchers who hope to have a closer examination on people’s sitting behavior. We hope that this observation and measurement
will serve as a prototype for further studies and applications that can provide us a foundation for understandingpeople’s sitting behaviors in open/public spaces. As mentioned, We will be able to further investigate on factors impacting the behavioral decision by broadening the scope and eliminating limitations we currently face for current project.