Video-Based HRI Studies

Description

Show participants filmed scenarios of a robot interacting with people and collect their reactions, instead of running costly live trials.

Live human-robot trials are expensive, slow and risky. In video-based HRI the team scripts and films a robot performing a scenario with people, often using Wizard of Oz control, and shows the videos to groups of participants, collecting reactions through questionnaires. Comparisons found live and video trials gave comparable results where a robot approached a person, but those scenarios involved little dynamic coordination. The more contingent and coordinated the interaction, the less video simulates live experience, so validity depends on the scenario. The method trades fidelity for scale and control: many viewers, identical stimuli and systematic variation of behaviours such as approach direction or speed.

Attributes

Loss type
Siloed
Stage
Design
Data
Mixed
Effort
Weeks

Steps

  1. Script scenarios that vary the robot behaviour of interest, such as approach direction or speed.
  2. Film the robot performing each scenario with people, using Wizard of Oz control if needed.
  3. Show the videos to groups of participants in a controlled order.
  4. Collect responses with questionnaires and optional discussion.
  5. Compare responses across behaviour variants.
  6. Validate key findings with live trials, especially for highly interactive behaviours.

Sources

  • HCI Encyclopedia (Soegaard & Dam, 2014)

Also known as

  • Video-based HRI (VHRI)
  • Video-based HRI methodology
  • Video trials

Categories

  • User Research
  • Usability & Evaluation