Embodied Stability in a Minimally-Actuated Soft Robot for Autonomous Exploration
Lior Salem, Adam Vichik, Amir Gat, Yizhar Or
AI summary
Problem
Conventional confined-space robots struggle with high control complexity and limited shape retention, while traditional soft robots lack passive stability and memory without sustained actuation.
Approach
The robot uses a serial chain of multi-stable elastic elements that passively lock into discrete shapes, with a single mobile pneumatic actuator triggering reversible transitions. Autonomy is achieved by integrating nonlinear hybrid modeling, visual pose estimation, and sampling-based motion planning within ROS2.
Key results
- Nonlinear hybrid-dynamics model integrated into a custom Gazebo plug-in for hardware-compatible simulation
- Novel on-board visual pose estimation system for accurate online configuration tracking
- Sampling-based motion planning framework tailored to multi-stable kinematics and feasible motion primitives
- Experimental validation of closed-loop autonomous navigation in cluttered environments with strong model-experiment correspondence
Why it matters
This mechanically intelligent architecture enables scalable, low-power reconfigurable robots for search-and-rescue, inspection, and medical applications where continuous actuation and complex control are impractical.
Abstract
Soft robots offer an opportunity to embed intelli- gence directly into morphology, potentially reducing the need for continuous feedback regulation. We present an autonomous, minimally actuated multi-stable soft robot for exploration in confined and cluttered environments. The robot is composed of a serial chain of multi-stable elastic elements whose energy land- scape encodes discrete, passively stable configurations, enabling reversible shape transformation and shape retention without sustained actuation. A single mobile pneumatic actuator triggers transitions between these stable states, producing complex three- dimensional configurations with minimal hardware complexity. Autonomy is achieved through the integration of nonlinear hybrid modeling, visual pose estimation, and sampling-based motion planning within a ROS2 framework. Rather than regulating continuous deformation, computation in our system selects and sequences mechanically admissible state transitions, while structural multi-stability provides inherent stabilization and memory. Experimental results demonstrate closed-loop navigation in cluttered environments using this distributed balance between mechanics and control. These results highlight an alternative organization of auton- omy in soft robotics, where feedback and planning operate over discrete embodied states while low-level stability is handled by the material and structural design.