AffectiveTouch - An experiment on pleasantness

Project Overview:

The pleasantness of an affective stroke is a very subjective thing, but science has a few explanations for the mechanisms at play. These mechanisms are different based on the type of stroke, which is why in this study, we look at the differences between two types of strokes, made on the forearm: one with a soft paint brush, and the other with vibrators which activate sequentially, called Apparent Haptic Motion. Specifically, we analyze the impact of the velocity of these strokes on the pleasantness rating given by the subjects. Through 3 different experiments, we isolate different variables to come to our final conclusion.

Objectives

  1. Make a robotic arm perform brush strokes on the lower arm in a precise manner, at arbitrary speeds.
  2. Create a smooth user experience for the participants and the researcher.
  3. Collect and analyze participant data.

Technology Stack

  • Frontend: Pygame for the user interface.
  • Backend: C++ for robot control and Python for experiment setup.
  • IPC: Redis in Docker.
  • Virtual Reality: Unity 3D.

Features

  1. Easy Researcher Interface

Researchers often need to manage their experiments manually, for example by collecting participant data in a spreadsheet or starting stimuli with a mouse and keyboard before each trial (which there can be dozens of). Doing this for hours in a day leads to fatigue, but also errors and inaccuracies in the data and stimuli. This project focuses on quality of life for researchers by adding the following features:

  • Automated startup script: very little knowledge needed to start this complex system. Just plug, turn on, and click start.

  • One-button experiment launch: at the click of a button, start the experiment for a given participant and handle all the necessary steps, like selecting the correct stimulus, playing given stimulus, waiting for the participant’s output, and pausing between each stimulus.

  • Automated data collection: the participant’s answers are collected in a file anonymously, and paired with the stimulus data for an easier analysis.

  • Pause, Fast-Forward, Replay: this flexibility allows the participant or the researcher to pause at any time during the experiment, replay a missed stimulus, or fast-forward in case of a rare error.

  1. Intuitive Participant Interface
  • An answer interface presents the required output modalities but lets the participant navigate freely. It is simple but intuitive, with special care to the user experience (experiments 1 & 2).
  1. Reliable Device Control

This project interfaces multiple complex components: the robotic arm, the vibrator controller, the user interfaces, and the input devices. All these parts have to work together and correctly.

  • Highly tested stimuli by: force feedback analysis, video analysis for stroke speed, accelerometers, …
  • Modular architecture centered around a message passing system.
  • Reliable software components that react to errors in a safe manner.
  1. Virtual Reality Environment
  • Immersive VR environment for experiment 3 where the participant can move their arms and hands realistically, using a neutral avatar, see the strokes being made, and give their output for each trial.

Outcome

Work-In-Progress published here, full research paper coming soon.

Over the more than 24 participants in the study, the system had a 90% reliability rate with very few errors and no unrecoverable errors. The participants and researcher were also pleased with the ease of use of the system. The automation level allowed the researcher and myself to pay more attention to the participants to improve the experiment and experiment conditions, as well as save time to work on other projects.