Arno Hartholt, Director, ICT’s Integrated Virtual Humans group, will present recent research developments in intelligent agent technologies at the ACM International Conference on Intelligent Virtual Agents (IVA), taking place in Berlin, Germany, from September 16 to 19, 2025.
The paper “Multi-Platform Intelligent Agents with the Virtual Human Toolkit” (Arno Hartholt, Ed Fast, Kevin Kim, Andrew Leeds, Edwin Sookiassian & Sharon Mozgai) details how ICT’s Virtual Human Toolkit supports the research and development of intelligent agents across multiple computing environments. The platform offers modular, extensible capabilities that integrate both academic and commercial technologies, facilitating experimentation and deployment on diverse hardware, including Windows & Mac desktops, mobile, and Quest 3 augmented/virtual reality headsets.
Key features of the platform include support for audio-visual sensing, automated speech recognition (ASR), natural language processing (NLP), text-to-speech synthesis (TTS), and nonverbal behavior generation. Demonstrations planned for the conference will highlight the system’s interoperability, flexibility, and ability to incorporate third-party character assets from tools such as Reallusion Character Creator and Headshot.
The updated platform addresses current trends in intelligent agent development, including increasing reliance on cloud services, expanded availability of commercial IVA-related technologies, and a broadening of supported hardware platforms beyond traditional desktop systems to mobile and XR devices.
Built on ICT’s real-time simulation environment RIDE, the platform maintains compatibility with standard behavior markup languages and offers application programming interfaces (APIs) that enable developers to select and integrate various technology implementations. This design allows for customization based on project-specific requirements and supports both local and cloud-based computing environments.
The platform includes a range of character assets, from legacy internally developed models to freely available and commercial third-party characters, which can be deployed across supported platforms.
ICT plans to release the platform’s source code publicly in Q4 2025 to promote wider adoption and collaborative development within the research community.
//