LuminAI is an interactive art installation in which participants canengage in collaborative movement improvisation with an artificiallyintelligent virtual dance partner.

The line between human and non-human is blurred, spurringparticipants to examine their relationship with AI based technologyand how it can be expressive, social, and playful.

LuminAI’s virtual agent analyzes participant movements through procedural representations of the Viewpoints andLaban’s movement theory (from theater and dance) and improvises responses from transformed memories of pastinteractions with people. In other words, the agent learns how to dance by dancing with us.

Viewpoints

Laban’s

The LuminAI installation ultimately examines how humans and machines can co-create experiences together asequals, but it does so in a lighthearted environment. Participants are able to creatively explore movement whilehaving fun.

ON-GOING WORK

Engaging with the public in aninformal learning environments,as a way of increasing theunderstanding and awareness ofAI and computational literacy.

Investigating the Laban’smovement theory as a wayfor the agent to understandand reason about movement

Developing a machine learningtoolkit for visualizing how theagent clusters similar gesturestogether as it is learning

Source: https://adamlab.gatech.edu/projects/luminai/

I am a Graduate Research Assistant in Dr. Brian Magerko’s Expressive Machinery Lab, working on the project. I amresponsible for imparting the learnings from Laban’s Movement Analysis to the AI. This includes:

Studying Laban’s MovementTheory to identify keys aspectsof human movement

Researching the variousadaptations of various LMAcomponents

Deriving computationalrepresentation of LMAcomponents being

Adopting the Laban basedcomputational modelsinto AI

Rudolf Laban, and later others, developed a method and language to describe, visualize, interpret and documenthuman movement, called Laban Movement Analysis. The method uses a multidisciplinary approach, incorporatingcontributions from anatomy, kinesiology, psychology, Labanotation and many other fields.

Source: Wikipedia

“LMA provides a rich overview of the scope of movement possibilities.These basic elements can be used for generating movement or fordescribing movement. They provide an inroad to understanding movementand for developing movement efficiency and expressiveness. Each humanbeing combines these movement factors in his/her own unique way andorganizes them to create phrases and relationships which reveal personal,artistic, or cultural style.”

Peggy Hackney, Making Connections

PARADIGMS

Effort

Space

Shape

Body

The way the bodychanges shape duringmovement is furtherexperienced andanalyzed through theShape category.

Describes the bodyparts that are moving,are connected, areinfluenced by others,and general statementabout organization.

Effort describes thedynamic qualities ofmovement. It isintended towardsunderstanding theinner intentionsrelated to a movement

Involves motion inconnection with theenvironment, and withspatial patterns,pathways, and lines ofspatial tension.

Source: Wikipedia

MOVEMENT QUALITIES

Efforts or Movement Qualities in Laban’s theory attempts to describe the dynamic qualities of human movement.These include qualities like how sudden a movement is, or how free a movement is, amongst others. Each of thesequalities are correspond to an inherent attitude, an innate nature. It is this innate nature that Laban tries to manifestthrough these qualities. There are 4 of these movement qualities which, in Laban’s Notation, are represented asfollows:

Weight

Space

Flow

Time

Laban’s notation for efforts

Time

Flow

Weight

Space

How heavy or light
the movement is.
Physiologically, it
relates to how much
energy is exerted in
movement

How free or bound the
movement is.
Physiologically it
relates to muscle
tension during the
movement

How direct or indirect
the movement is. This
relates to how focussed
the attention of a
performer is

How sudden or
sustained the movement
is. This is related to how
jerky the motion is

ATTENTION

INTENTION

DECISION

PRECISION

You attend to an object

You identify the
intention of dealing
with object

You choose, quickly or
slowly, whether to
pursue the decision

You carry out the
intended action with
certain precision

The virtual dance partner manifests itself as a shadow on thescreen in front of participants. This shadow tries to follow themovements of the participant - starting with imitation and thengoing into response mode. In response mode, the agent can doeither do random transformation on the same movement asthe participant, or recall a similar gesture from its memory.

The gesture recall is based on recognition of the current movement by the agent and finding another gesturewith physical similarity or temporal closeness. The agent keeps building and refining its knowledge base based oninteraction with the participant.

ARCHITECTURE

Movement
Theory

Gesture
Segmentation

LuminAI Core

Decision
Making

Visualization

Memory

Currently, we are working on adopting the ‘Time’parameter of Laban’s Efforts into the system. This is doneby capturing joint velocity and acceleration, and usingthese to identify sudden and sustained movements.

The next steps involve adoption of ‘Space’ and ‘Weight’qualities of Effort using multi-modal input like MotionCapture suit, Kinect, EMG Band, Accelerometer etc. Wewill also plan on exploring spacial qualities of dance andmovement using the icosahedron-based kinesphereconcept of shape and space exploration.