Re-Purposing a Modular Origami Manipulator Into an Adaptive Physical Computer for Machine Learning and Robotic...

Area of Science:

• Robotics and Mechanical Engineering
• Computational Science
• Materials Science

Background:

• Physical computing offers an alternative to traditional CMOS computers for intelligent tasks in mechanical systems.
• A systematic understanding of how mechanical design influences physical computing performance is lacking.

Purpose of the Study:

• To investigate the relationship between mechanical design and physical computing performance.
• To evaluate the computing capacity of an adaptive physical reservoir based on its configuration.
• To demonstrate the potential for embodied intelligence in soft robots and functional materials.

Main Methods:

• Repurposed an origami-inspired modular robotic manipulator as an adaptive physical reservoir.
• Systematically evaluated computing capacity with varying physical configurations, input setups, and tasks.
• Utilized the NARMA benchmark for time series emulation and analyzed nodal dynamics for perception capabilities.
• Integrated shape memory alloy (SMA) actuation for robotic operations.

Main Results:

• Time series emulation performance directly correlates with the Peak Similarity Index (PSI), reflecting reservoir dynamics.
• The adaptive reservoir demonstrated perception by extracting payload weight and orientation from intrinsic dynamics.
• Information extraction capability is measurable via spatial correlation between nodal dynamics.
• SMA actuation enabled practical robotic operations by exploiting embodied computing power.

Conclusions:

• A strategic framework for harvesting computing power from soft robots and functional materials was established.
• Design parameters and input selection can be configured based on specific computing task requirements.
• This approach enables next-generation embodied intelligence in bio-inspired materials and robotic systems.