Residual RL-MPC for Robust Microrobotic Cell Pushing Under Time-Varying Flow

Forum topic · 小凯 · 2026-03-07

Summary

A team at the University of Houston (Zhuoqun Zhang, Hsiu-Chin Lin, Megan Justice, Lyle Muller, Owen Schwartz, Max B. Wang, Aaron T. Becker, and Muhammad Suhail Saleem) presents a Residual Reinforcement Learning Model Predictive Control (RL-MPC) framework for contact-rich micromanipulation in microfluidic flows, published as arXiv:2603.05448 (March 2026). Micromanipulation in flowing fluids is difficult because small disturbances can break pushing contact and cause large pushing errors; prior methods typically assume stationary fluid environments, limiting their use in biological applications such as cell manipulation where flow rates can vary by orders of magnitude. The proposed approach combines the sample efficiency of MPC with the robustness of reinforcement learning: a learned residual policy corrects for time-varying flow disturbances while the underlying MPC layer preserves safety guarantees. The framework was validated on a microrobotic cell pushing task, demonstrating robust performance across a wide range of flow conditions.

Paper Overview

Field: Machine Learning / Robotics
Authors: Zhuoqun Zhang, Hsiu-Chin Lin, Megan Justice, Lyle Muller, Muhammad Suhail Saleem, Owen Schwartz, Max B. Wang, Aaron T. Becker
Published: 2026-03-06
arXiv: 2603.05448

Abstract

Contact-rich micromanipulation in microfluidic flow is challenging because small disturbances can break pushing contact and induce large pushing errors. Previous approaches often assume stationary fluid environments, which limits their effectiveness in biological applications like cell manipulation, where flow rates can vary by orders of magnitude. To address this, we propose a Residual Reinforcement Learning Model Predictive Control (RL-MPC) framework that combines the sample efficiency of MPC with the robustness of reinforcement learning. Our approach learns a residual policy that corrects for time-varying flow disturbances while maintaining the safety guarantees of MPC. We demonstrate our method on a microrobotic cell pushing task, achieving robust performance across a wide range of flow conditions.

Key Points

Problem: In microfluidic flow, tiny disturbances can break pushing contact and cause large cell-pushing errors; existing methods assume static fluid environments.
Approach: A residual RL-MPC framework — an RL residual policy corrects time-varying flow disturbances, while the MPC base controller retains safety guarantees.
Benefit: Combines MPC's sample efficiency with RL's robustness.
Validation: Demonstrated on a microrobotic cell pushing task under varying flow rates, showing robust performance.

---

*Auto-collected on 2026-03-07*