Within SHARESPACE, cognitive architectures were designed, developed, and validated to drive the movement of virtual characters with different levels of autonomization. Two main kinds of architectures were developed. Those to increase synchronization in a group and those to amplify kinematic information in human motion. These architectures were thus designed for different use cases, and deployed both in the project’s proof-of-principle demonstrations and the application scenarios
The methodologies developed demonstrate that the Cognitive Architectures can, in the case of L2 virtual characters, correctly and informatively amplify human motion, making kinematic information easier to interpret. For L3 virtual characters, they enable integration into a group, support synchronization, and ensure seamless blending within the ensemble. These techniques have been evaluated and refined within the project’s application scenarios.
The design principles, algorithmic derivations, and technical specifications of these cognitive architectures are further described in the associated scientific publications. All software components were developed in Python and released as open-source. In particular, the project produced the following software packages:
CA for Health, CA for Popsync, CA for Amp, CA for Art, CA for sports
- For use in the Proof of Principle of Amplification: https://github.com/FrancescoDeLellis/L2_Cognitive_Architecture_PoP_Amplification
- For use in the Proof of Principle of Synchronization:
https://github.com/SINCROgroup/Cognitive-Architecture-for-Synchronization - For use in the Health Scenario:
https://github.com/SINCROgroup/Cognitive-Architecture-for-Health - A standalone integrated module for the phase estimation of pseudo-periodic signals
https://github.com/SINCROgroup/recursive-online-phase-estimator - For use in the Art Scenario:
https://github.com/FrancescoDeLellis/L3_cognitive_architecture/tree/main/CA_falcon_heavy
