A Unified Control Architecture for Macro-Micro Manipulation using a Active Remote Center of Compliance for Manufacturing Applications

Macro-micro manipulators combine a macro manipulator with a large workspace, such as an industrial robot, with a lightweight, high-bandwidth micro manipulator. This enables highly dynamic interaction control while preserving the wide workspace of the robot. Traditionally, position control is assigned to the macro manipulator, while the micro manipulator handles the interaction with the environment, limiting the achievable interaction control bandwidth. To solve this, we propose a novel control architecture that incorporates the macro manipulator into the active interaction control. This leads to a increase in control bandwidth by a factor of 2.1 compared to the state of the art architecture, based on the leader-follower approach and factor 12.5 compared to traditional robot-based force control. Further we propose surrogate models for a more efficient controller design and easy adaptation to hardware changes. We validate our approach by comparing it against the other control schemes in different experiments, like collision with an object, following a force trajectory and industrial assembly tasks.

💡 Research Summary

The paper addresses a fundamental limitation in macro‑micro manipulation systems used for high‑speed, high‑precision manufacturing tasks. Conventional architectures assign position control to the macro robot (large‑workspace industrial arm) and force control to a lightweight micro end‑effector, which restricts the overall interaction bandwidth because the macro robot’s inertia is not part of the active force loop. To overcome this, the authors propose a unified control architecture that integrates both the macro manipulator and the micro active end‑effector (an Active Remote Center of Compliance, ARCC) into a single active force‑control loop.

The system is modeled as a mass‑spring‑damper network. The macro robot’s dynamics are approximated by a second‑order low‑pass transfer function G_M(s) = K_M·ω_cM²/(s²+2ζ_M ω_cM s+ω_cM²). The ARCC’s active side is similarly modeled by G_μa(s) = K_μa·ω_cμa²/(s²+2ζ_μa ω_cμa s+ω_cμa²). The passive side, which includes the flexure hinge and the environment stiffness k_env, is described by G_μp(s) = (c_μ s + k_μ)/(m̃_μp s² + c_μ s + (k_μ + k_env)). The overall micro‑manipulator transfer function is the cascade G_μ(s) = G_μp(s)·(1/s)·G_μa(s). Parameters are identified using sine‑sweep excitation and an instrument‑variable algorithm. Identified macro cutoff frequencies are around 3 Hz with a damping ratio near unity, while the micro active side exhibits cutoff frequencies of 20–27 Hz and damping ratios of 0.4–0.45, confirming a much higher bandwidth.

Virtual springs and dampers are introduced to shape the closed‑loop dynamics. For the macro robot, a virtual damper c_ctrl,F,M converts the force error (desired minus measured contact force) into a velocity command: G_F,M(s) = 1/c_ctrl,F,M. The micro side uses an analogous damper c_ctrl,F,μ for force control, and a PDT1‑type position controller G_x,μ(s) = (k_ctrl,x + c_ctrl,x·s)·T_filter·s/(1+T_filter·s) to keep the micro end‑effector centered in its workspace. This results in two parallel admittance‑type force loops (macro and micro) and a position loop for the micro, all operating simultaneously.

Controller synthesis employs fixed‑structure H∞ optimization (Apkarian, Bruinsma). The generalized state‑space model T(s) includes the identified plant dynamics and tunable controller blocks. Performance weights are defined for force tracking error (p_F), position error (p_x), and velocity error (p_ẋ). The optimization minimizes these weighted norms while guaranteeing robustness to disturbances w and measurement noise.

Experimental validation comprises three scenarios: (1) a sudden collision with a rigid object, (2) tracking a prescribed force trajectory, and (3) an industrial assembly task involving peg‑in‑hole and contour following. Compared with the state‑of‑the‑art leader‑follower (LF) scheme, the proposed architecture achieves a 2.1× increase in interaction bandwidth; compared with a traditional robot‑based force control (RB) it achieves a 12.5× increase. Force overshoot and vibration are significantly reduced, leading to smoother, more stable contact. Moreover, the surrogate model approach enables rapid retuning when hardware changes (different robot or ARCC module) by simply re‑identifying a few parameters, demonstrating modularity and practical adaptability.

Key contributions are: (i) a novel unified control architecture that incorporates both macro and micro manipulators into the active force loop, delivering a substantial bandwidth boost, (ii) derivation and identification of surrogate dynamic models that facilitate efficient controller design and hardware‑agnostic adaptation, and (iii) thorough experimental comparison against LF and RB baselines across realistic manufacturing tasks. This work advances the feasibility of high‑performance macro‑micro robotic systems for dynamic physical interaction in modern manufacturing environments.