Success! Thanks for contacting us.

Beyond Imitation

'I didn’t want to imitate anybody. Any movement I knew, I didn’t want to use.' – Pina Bausch

Recent advancements in motion capture and machine learning have created provocative new possibilities in the domain of movement generation. This project fuses expertise in machine learning, dance practice, and particle physics to create open-source tools for the research of experimental dance.

Encoding Motion

We have created a suite of publicly-accessible tools for generating novel sequences of choreography as well as tunable variations on input choreographic sequences using recurrent neural network and autoencoder architectures. These methods have been developed using improvisational dance from our team members, recorded using a state-of-the-art motion capture system with a rich density of over 50 datapoints representing the human form.

See the Code

Photo Credit: Jessica Todd Harper

Our Mission

We develop machine learning-based tools to inform movement-based research. We believe that technology, in conjunction with physical practice, can enhance embodied learning. We are equally interested in ethical, aesthetic, and technical questions. We aim to make kinesthetic experimentation accessible to all.