MicroPush: A Simulator and Benchmark for Contact-Rich Cell Pushing and Assembly with a Magnetic Rolling Microrobot
New simulator enables reproducible testing of AI-controlled microrobots for delicate cell assembly tasks.
Researchers Yanda Yang and Sambeeta Das have introduced MicroPush, a new open-source simulator and benchmark suite designed specifically for testing magnetic rolling microrobots in microscale manipulation tasks. Published on arXiv, this tool addresses a critical gap in robotics research by providing a standardized environment for developing and evaluating autonomous control systems for contact-rich behaviors like cell pushing and multi-target assembly in cluttered microfluidic environments. Unlike physical experiments that are difficult to reproduce, MicroPush offers researchers a consistent platform to test planning algorithms, control strategies, and machine learning approaches for delicate operations at the cellular level.
The simulator combines an overdamped interaction model with realistic physics including contact-aware stick-slip effects, near-field damping, and optional background fluid flow (Poiseuille flow). It features a calibrated mapping from magnetic actuation frequency to robot rolling speed and comes with a modular two-phase planning-control stack for contact establishment and goal-directed pushing. The benchmark protocol includes fixed tasks for single-object transport and hexagonal assembly with unified CSV logging, measuring success rates, completion time, tracking accuracy, and actuation variation (EΔω). Early results show controller stability is crucial under flow disturbances, while planner choice affects command smoothness in long-horizon sequences. This tool enables reproducible comparison and ablation studies that could accelerate development of autonomous microrobots for biomedical applications.
- Open-source simulator models magnetic microrobot physics with stick-slip contact and fluid flow effects
- Includes benchmark suite with fixed tasks for single-cell transport and hexagonal multi-target assembly
- Provides modular planning-control stack and standardized metrics (success rate, time, EΔω) for reproducible AI testing
Why It Matters
Accelerates development of AI-controlled microrobots for precise biomedical tasks like cell sorting and tissue engineering without costly physical experiments.