Skip to main content

intro

Human-Robot Interaction (HRI) Consortium – 3-Year Activity Overview​

Over the past three years, the HRI Consortium—comprised of Israel’s leading academic institutions (Bar-Ilan University, Ben-Gurion University, Tel Aviv University, the Technion, and Reichman University) and prominent robotics and AI companies (Cogniteam, Robotican, RGo Robotics, Intuition Robotics, ReWalk Robotics, and Elbit Systems)—has spearheaded a national effort to advance Human-Robot Interaction technologies. This initiative was supported by a $15 million grant from the Israel Innovation Authority.

At the heart of this initiative is the HRI Toolkit, an open, smart, and modular software architecture designed to enable seamless, socially aware interactions between robots (or smart edge devices) and humans. The development of this toolkit was led by Cogniteam and represents a significant technological leap forward in the domain of cognitive robotics and human-centered AI.

Key Attributes of the HRI Toolkit​

Platform-Agnostic and Modular​

Built on the ROS2 (Robot Operating System), the toolkit integrates easily with any robot or smart device capable of exposing sensory data streams and actuation APIs. This flexibility allows for rapid deployment across various platforms and applications.

Cognition-Driven Architecture​

The HRI Toolkit is structured into multiple cognitive layers:

  • Human Context Layer: Detects, tracks, and interprets human behaviors and social cues through visual and vocal inputs.
  • Scene Context Layer: Maps and understands the surrounding environment, including object recognition and SLAM.
  • Context Management Layer: Synthesizes contextual insights, detects contradictions or anomalies, and classifies social contexts.
  • Social Planning Layer: Plans context-aware, socially appropriate actions such as navigation, speech adaptation, and help requests.
  • Actuation Layer: Executes responses naturally through speech, movement, and expressive gestures.

High Configurability and Extensibility​

The architecture distinguishes between:

  • Stable Framework Components (e.g., skeleton detection, data fusion)
  • Rapidly Evolving Capabilities (e.g., gesture detection, speech-to-text)

New technologies and models can be integrated via simple JSON-based configuration, ensuring futureproofing and rapid technological adaptation.

Transparent and Safe Reasoning​

The consortium adopted a hybrid AI approach:

  • Core decision-making is based on classical, explainable AI methods.
  • LLMs (Large Language Models) are used for narrow, controlled functions (e.g., speech rephrasing), but not for critical decision-making.

This ensures robust, transparent, and safe AI behavior, with all data openly accessible in an internal knowledge base.

Broader Impact​

Beyond robotics, the HRI Toolkit paves the way for transformation across multiple human-facing industries—such as eldercare, healthcare, education, and customer service—by enabling machines and smart devices to interact more naturally and effectively with people. It lowers the barrier to entry for developing socially aware, context-sensitive systems.

The HRI Consortium's work represents a major advancement in the development of responsible, explainable, and human-centric AI systems, with the potential to significantly impact how humans and intelligent systems collaborate in the years to come.