A research team from Fraunhofer HNFIZ has introduced a new evaluation model that ranks humanoid robots across five levels of technical capability. It also categorizes applications by the abilities they require, enabling easier comparison between humanoids, helping match the right robot to each use case, and identifying gaps in current technology development.
Amid a growing shortage of skilled workers and rapid progress in artificial intelligence (AI) and robotics, humanoid robots are gaining attention in industry and logistics. Companies are now deciding when and where to deploy these robots in a technically and economically viable way. While many businesses and research institutions are already testing early systems, a comprehensive and standardized evaluation model has been missing until now.
Clearer Insight and Easier Comparison for Humanoid Robots
This is where the Humanoid Capabilities Navigator comes in—a model developed by the Fraunhofer Institute for Manufacturing Engineering and Automation IPA at its Heilbronn site and presented in a recent white paper. As part of the Fraunhofer Heilbronn Research and Innovation Centers, the institute’s AI-based Robotics Research and Innovation Center also explores humanoid robotics alongside intelligent mobile manipulation and the flexible use of robotic hands.
Inspired by the five levels of automation in autonomous driving, the Humanoid Capabilities Navigator assesses humanoid robots across four broad categories, each covering multiple technical capabilities and stages of development:
- Mobility and movement (e.g., localization and path planning)
- Manipulation (e.g., force sensing and grasp planning)
- Cognition (e.g., perception and task planning)
- Safety and security
Within each category, capabilities are rated on maturity levels from 0 (absent) to 4 (highly autonomous, human-level or beyond). This allows, for the first time, a transparent, vendor-neutral, and robust comparison of humanoid robots, while also indicating how well they fit specific applications.
Real-World Applications and Robots Classified in Practical Terms
The publication demonstrates how the Humanoid Capabilities Navigator can be applied to common industrial tasks such as truck loading and unloading, order picking, machine operation, and maintenance. For each use case, it outlines the required maturity levels across the four capability areas. For instance, truck loading demands advanced manipulation, cognition, and safety, while machine operation requires lower capability levels due to its more structured and predictable nature.
The automation division of Fraunhofer Institute for Manufacturing Engineering and Automation IPA also tested the model using a current humanoid robot, the Unitree G1 humanoid robot from Unitree Robotics. The results show that the G1 has already made notable progress in areas like mobility, but still requires significant development to fully meet industrial demands. The evaluation framework clearly highlights these gaps and helps guide future development priorities.
Guiding Research, Development, and Market Projections
With the Humanoid Capabilities Navigator, Fraunhofer Institute for Manufacturing Engineering and Automation IPA provides industry with a practical tool to realistically evaluate the potential of humanoid robots, inform investment decisions, and guide targeted research and development efforts. It also allows for a reliable projection of future market trends by systematically linking technological progress to specific application needs.
The white paper is part of a series that the institute has already published or plans to release in the first half of 2026, all aimed at giving companies thorough and dependable resources to assess the technical and economic feasibility of humanoid robots for their own use cases.
Read the original article on: Tech Xplore
Read more: BMW Factory Deploys humanoid robots powered by physical AI