Perkins + Will

 

I worked at Perkins + Will from 2017-2018. Here are some of my favorite projects I got to work on.

 
 

Avatarworld

Avatarworld was an internal product at the Chicago Perkins+Will. It used object detection to extract context as to what is happening in a space, and gave people a UI to navigate rooms virtually. There was an interesting moment where the boundaries of privacy in an open office where question and lead to a few great discussions among leadership, in reference to the future of workplace.

 
 

  • 2019

  • I ran YOLO v3 for Real-Time Object Detection at the edge and pushed the data to AWS. The visualization was built in Unity 3D. The Avatarworld platform was deployed on the web.

  • People thought it was awesome to visualize the vibe of spaces through a big-brother-like experience, however they were a bit creeped out by accuracy of the real-time render. It was important for us to think about the boundaries of private spaces in offices as technology begins to blur that boundary through IoT and ubiquitous computing.

 
 
 
 
 

Gesture Detection

We believe that in the future, we will use gesture-based interactions to communicate with machines in spaces. This exploration uses computer vision to detect counts of thumbs up interactions. Through these ubiquitous forms of surveys, we can understand space in a new way and provide solutions to problems occurring in real-time.

 
 

  • 2019

  • This project runs at the edge using a Raspberry Pi. The software is written in Python using Yolo V2.

  • People felt positive around giving constructive feedback to facilities and shared services in the office. Often compared to the smiley face surveys in airports and places like Ikea, people felt that removing the interface may normalize people’s involvement in their work community.

 
 
 
 
 

I created a hardware device to help me train the gesture detection model. This new piece of hardware made the training process quicker and allowed me to train the model on other gestures.

 
 

CubeIO

CubeIO is an IoT demonstration in the form of a digital human interface device to allow for real-time polling during meetings, without the use (and distraction) of the user's personal smart device.

We believe that an intuitive, simple and physical interaction keeps the user focus on the question, their thoughts, and the responses of the group. The equity of all voices anonymously participating drives decision making while also creating a shared experience for everyone in the room.

 
 

  • 2018

  • CubeIO is custom hardware built with the Particle IoT system. I created the PCB in Eagle and laser cut the first iteration of the design. The second iteration was 3d printed at an industrial facility. The firmware was written in C++.

  • CubeIO was primarily used in C-level meetings every half. People felt that it was really fun to use the cubes and encouraged a playful way to mediate debates.