Hierarchical Multimodal Motion Control of Magnetic Pivot-Walking Millirobotic-Grippers for Autonomous Target Acquisition in Complex Terrains

Magnetic soft millirobotic-grippers, equipped with agile pivot-walking motions and adaptive enveloping morph abilities, hold great promise for biomedical target acquisition tasks. However, deploying these millirobotic-grippers in highly constrained, disturbed, and variable terrains to realize multisequence target acquisition and transport tasks remains challenging. Here, we introduce a hierarchical multimodal motion control method for pivot-walking magnetic milliroboticgrippers, which enhances adaptive locomotion capabilities and enables high-precision motion control, facilitating autonomous target acquisition in complex terrains. The millirobotic-gripper utilizes a centrosymmetric three-pivot design, enabling adaptive soft enveloping deformation and robust multimodal locomotion. A hierarchical control architecture is proposed, comprising: 1) an upper-level Event-Based Finite State Machine planner that dynamically orchestrates transitions between motion modes according to environmental feedback and task-specific conditions; and 2) a lower-level Sliding Mode controller integrated with Gaussian Process-based gait parameter optimization, significantly improving motion accuracy and robustness against environmental disturbances. Experimental results demonstrate that our proposed method allows millirobotic-grippers to efficiently navigate morphing tunnels, leap across gaps exceeding 3 times their body length, accurately follow arbitrary paths with errors less than 5% of their body length, and reliably perform three types of targets grasping and transport. Furthermore, the biomedical application potential of our system is initially validated through ex vivo porcine gastrointestinal tract experiment with ultrasound guidance.