2026 Theses Doctoral

Self-Guided Navigation of Magnetic Microrobots via Programmable Taxis in Time-Varying Fields

Wu, Yiyang

Microrobots operating at the scale of living cells hold promise for biomedicine, environmental sensing, and materials science, yet steering individual particles through complex environments remains a fundamental challenge. Because spatially uniform magnetic fields drive all particles identically, conventional feedback control cannot independently guide multiple particles of unknown position.

This dissertation develops a physical intelligence framework in which time-varying, spatially uniform magnetic fields encode self-guided navigation of ferromagnetic microspheres along local environmental gradients without external position feedback. Chapter 1 reviews magnetic actuation and self-guided navigation, surveying three forms of gradient-driven taxis (topotaxis, rheotaxis, and viscotaxis) that constitute the core of this work.

Chapter 2 reports topotaxis experiments in which toggled rotating fields drive ferromagnetic Janus spheres on inclined substrates: cross-slope velocity scales linearly with field frequency while along-slope speed grows quadratically, enabling uphill climbing at high frequencies. The findings extend to concave and convex topographies, and to multi-particle settings in which independent particles navigate under the same global field.

Chapter 3 develops a model for programmable rheotaxis by coupling magnetic torques to low-Reynolds-number hydrodynamics in the sphere–wall geometry. Tuning the frequency, magnitude, and waveform of the applied field selects among upstream, downstream, and cross-stream migration; evolutionary optimization identifies high-performance field protocols and exposes a fundamental trade-off between rheotactic speed and robustness to parameter uncertainty.

Chapter 4 presents a theoretical framework for viscotaxis grounded in the force–torque cross-coupling that emerges when a sphere rotates in a spatial viscosity gradient. The drift velocity, perpendicular to both the gradient and the rotation axis, can be reversed by switching the field handedness. To support experimental validation, a microchannel platform is developed that generates stable, controlled viscosity gradients while suppressing buoyancy-driven convection, enabling systematic characterization of the local viscosity profile.

Chapter 5 synthesizes these results and charts future directions: experimental validation of viscotaxis, non-spherical and three-dimensional extensions, navigation in non-Newtonian and pulsatile biological flows, and multi-cue operation in environments with simultaneous topographic, shear, and viscosity gradients.

Collectively, these results show that physical intelligence encoded in field waveforms alone is sufficient to navigate magnetic colloids through topographic, fluidic, and rheological complexity, establishing a new paradigm for microrobot control that scales naturally to swarms without sensors, feedback, or onboard computation.