Optimization and control of actuator networks in variable geometry truss systems using genetic algorithms.

Area of Science:

• Robotics
• Morphological Computation
• Bio-inspired Engineering

Background:

• Robot morphology is crucial for functionality, mimicking biological shape adaptation.
• Variable Geometry Trusses offer morphing capabilities but face control complexity challenges.
• Existing metatruss designs require manual control network configuration.

Purpose of the Study:

• To automate the design of control networks for metatrusses.
• To optimize actuator grouping, contraction ratios, and actuation timing for complex shape adaptations.
• To reduce control complexity in morphing robotic systems.

Main Methods:

• Developed a multi-objective optimization framework using a tailored genetic algorithm.
• Created a highly damped dynamic simulator balancing computational efficiency and physical accuracy.
• Validated the approach through experimental prototypes across multiple tasks.

Main Results:

• Successfully automated actuator grouping and control sequences for metatrusses.
• Demonstrated complex shape adaptations with significantly reduced control units.
• Identified an optimal number of control networks for performance gains.

Conclusions:

• The automated framework enables efficient and complex shape adaptations in morphing robots.
• Optimized metatrusses offer a viable solution for advanced robotic morphology.
• The study highlights the trade-off between control network number and performance benefits.