#habitualinstinct

The meeting with my primary advisor, Nick Puckett yesterday went well. I showed him the progress I’ve made with the first iteration of my thesis, Habitual Instinct. We discussed some of the issues I was having technically as well as construction of the current object, next iteration of movement as well as the collection, storage and transfer of data collection to a Processing of openFrameworks app.

Todo

Fix measurement accuracy

Fix interaction speed

Create a data structure to be passed to a 3d app for visualizing what the object is doing

Work on basic visualization of how the measurements the sensors take will be viewed

Add a secondary sensor. Potentially offset the movement it does with the other sensor. The movement is a mirrored opposite. For instance when one sensor is facing up, the other sensor is facing down.

Improve the solution that connects the sensor to the rotating axle. Use a small “L” bracket and use two nuts to hold it to the axle. Connect the sensor to the “L” bracket plate.

Concepts and Ideas

With multiple sensors, could create offset movement like a sine wave or other more complicated visuals with the sensors themselves.

Use the sensors to scan a full 180 or 360 degrees

Visualize the data points collected via color or 3d model

Try scanning a controlled space for an hour or more and see if there are any unique digital artifacts in the measurements

With a 360 degree scan, the output 3d model or output object could be an interesting take on a cylinder

Think of ways to use the distance measurements as well as timestamp plus the angle of the measurement

While working on the Habitual Instinct project I learned a lot in all aspects of the project ranging from the technical, production and how a piece is situated within its environmental. Some of the issues and learnings were more frustrating to deal with and figure out than others while others sparked an expansion of the original concept. The interaction with Habitual Instinct worked as I hoped in terms of the piece scanned the area in front of it, interpreted the data that was collected and then reacted to the the physical environment that it was trying to understand. The interaction by Habitual Instinct was to find the position with the largest measurement distance and orientate its sensor in that direction. Once pointing in the direction with the furthest distance it would continue to scan in that position until the distance changes. The distance would change by a the movement of a user within the exhibition space walking in front of the now static sensor. When the distance changes, it triggers the rescan by Habitual Instinct to try and find a new position with the largest measurement.

The main issue that I had with the creation of Habitual Instinct was getting consistently accurate reading from the sensor. To get more accurate readings I had to slow down the frequency of the stepper motor movement. This slow paced interaction interfered with the fluidity and user experience making it difficult to understand what exactly is happening with Habitual Instinct. I suspect the issue with accuracy was caused by sending and receiving pings that are not synced with the control movement of the stepper motor rotating the direction of the sensor. To solve this I am going to refactor the code to use interrupts by the sensor to trigger the movement of the motor. This will hopefully capture more accurate measurements as well as allowing for the interaction to be more smooth and more quickly. this was was I could only get it to read the measurements accurately when there was a noticeably large delay between the movements of the sensor. This made the user experience a little wonky. I can see how this would create confusion with the interaction and

Learnings

Controlling stepper motors and being aware what angles the sensor is pointing

External objects in the environment while testing will interfere with readings -- especially hollow or skeletal objects like chandeliers hanging from the ceiling mess up consistency of readings.

Responsiveness to interactions and change in sensor readings can differ when can change and be different than expected between distances under 100 cm and below 400 cm.

Issues

Getting accurate readings from the sensor while it is moving

Unhappy with the documentation / not in a gallery space

The documentation isn’t demonstrating the “full experience” due to space restrictions while filming.

New Steps

The movement for scanning needs to be more fluid and happen quicker

Responsiveness to human interactions needs to happen in real-time to offer the full effect.

To make it more believable and easier to understand, scanning and tracking the measurements need to happen when the sensor is rotating up and down.

Next Steps Recently i’ve been inspired by the works of Quayola and Joanie Lemercier. I find their representations of geometric shapes, and representing our reality and environment in the physical world by deconstructing objects into simpler shapes and displayed via a 3d modeling mesh which gets combined with the original object really interesting. I have also was curious about generative landscapes and how they are programmed. I found a great tutorial by Daniel Shiffman about 3D Terrain Generation. After learning a bit more about 3d meshes I started to think about the connection between the computers interpretation of what it sees and what is in reality there. With Habitual Instinct I am only measuring 14 different directions and distance measurements but there could potentially be millions of measurements taken if a realistic scene was needed to be accurately represented in the 3d space.

Thinking more about computer interpretation of the physical world I was thinking that for each pass Habitual Instinct makes, the measurements could be recorded and passed to a Processing app which will be added it to an arrayList which gets pushed into a 3d mesh. This mesh would be used to generate a topographical 3d plane to visualize the and better understand what the computer sees. The measurements gathered by Habitual Instinct will also display how data is gathered and stored online. An interaction with Habitual Instinct doesn’t happen one and is forgotten about, the interaction is added to an indefinite log of data which could be referenced in the future.

With this set of data collected by Habitual Instinct I was thinking it could be taken one step further by turning the mesh into a 3d model which could be milled out of a plank of 2” x 6” x 10’ wood or something similar. This would be interesting because it takes a digital measurement and transforms interactions that happen in the past and as a memory into a physical representation of that specific period in time. This could better demonstrate the physicality of how computers and sensors interpret their surroundings as well as the lasting storage of data that is gathered through technology but stored, preserved and used for alternative means.

Here’s a video of my first working prototype of an idea I had around avoidant technology and what happens when our expectations of technological behaviour is not what we assume. Confidence scans the area in front of it and focuses on the area with the most available space. If that space becomes blocked and the distance changes, ie. someone walks in front of the piece the object will rescan looking for more free space.