Active Constraints for Simultaneous Guidance and Safety in Handheld Micromanipulation

Roger Ho, Robert A. MacLachlan, Cameron N. RivierearXiv 2026 · arXiv

TL;DR

Researchers have developed a new optimization-based framework for handheld surgical robots that provides both task guidance and safety enforcement simultaneously. By using a "guidance degradation" strategy, the system ensures safety is never compromised while gracefully signaling to the surgeon when they need to adjust their hand position to stay within the robot's limited workspace.

What Problem Does This Paper Address?

In the world of microsurgery (like operating on delicate retinal veins), human hands are inherently limited by physiological tremors and tiny drifts. Handheld robots, like the "Micron" device used here, help by counteracting these tremors. However, providing "active constraints"—digital boundaries that guide a tool along a path or keep it out of a "forbidden zone"—is much harder for handheld devices than for large, floor-mounted robotic arms.

Handheld robots have two major hurdles:

  1. No Mechanical Ground: They actuate relative to the surgeon's hand, not a fixed base.
  2. Tiny Workspaces: They only have a few millimeters of "stroke." If a surgeon’s hand drifts too far, the robot hits its mechanical limit and can no longer help.

Existing handheld systems could usually only enforce one constraint at a time. If you turned on a safety constraint to avoid a delicate structure, you often had to turn off the guidance helping you hit your target, leading to a significant drop in accuracy.

Methods

The authors propose a framework based on Quadratic Programming (QP). This mathematical approach allows the robot to balance multiple goals every millisecond:

  • Safety as Hard Constraints: Safety requirements (like staying within a specific entry point, or Remote Center of Motion) are treated as non-negotiable. The authors use Control Barrier Functions (CBFs) to ensure the robot never violates these boundaries, even accounting for predicted hand motion.
  • Guidance as Soft Objectives: Task goals (like keeping a needle tip on a specific target) are placed in the "objective function." The robot tries its best to meet these goals using whatever "actuation budget" (stroke) is left after safety is guaranteed.
  • Guidance Weight Normalization: A novel technique that ensures the "feel" or bandwidth of the guidance remains consistent, regardless of the robot's current pose or configuration.
  • Guidance Degradation: This is the "secret sauce." When the robot runs out of room to satisfy both safety and guidance, it doesn't just fail or jerk. Instead, it gradually reduces the guidance, effectively "drifting" away from the target. This serves as a haptic cue to the surgeon to move their hand back toward a better centered position.
The experimental setup involves a 6-DoF handheld robot performing a needle-pointing task through a small hole (the safety constraint).

Datasets

The framework was evaluated using a custom experimental setup rather than a standard software dataset.

  • Hardware: A 6-Degree-of-Freedom (DoF) parallel kinematic handheld micromanipulator.
  • Task: Four novice subjects performed a "needle-hold" task for 330 seconds. They had to keep a needle tip on a virtual target while passing through a physical hole (simulating a surgical incision) that acted as a Remote Center of Motion (RCM) safety constraint.
  • Baselines: The system was compared against "Unaided" (robot off), "Shelving" (a standard tremor-canceling algorithm), and "AC-S" (Active Constraints for Safety only).

Key Insights and Ideas

  • The Safety-Accuracy Tradeoff: Enforcing safety alone actually decreases accuracy. Without guidance, the robot's "safest" pose often pulls the tip away from the target. By enabling simultaneous guidance, the robot can use its remaining degrees of freedom to satisfy both.
  • Degradation as Communication: Instead of safety being a "binary" failure, the gradual loss of guidance (seen in Figure 3) acts as a signal. Surgeons naturally "gravitated" toward the boundary where guidance begins to degrade, using it to keep their hand centered.
  • Feedforward Safety: Unlike grounded robots, handheld systems must account for the fact that the "base" (the hand) is constantly moving. The framework includes a feedforward term to counteract predicted hand motion before it causes a safety violation.
Graphs showing how tip error increases proportionally as the guidance fraction (alpha) decreases to preserve safety.

Key Results

The proposed framework (AC-S+G) showed dramatic improvements over the baselines:

  • Safety: Reduced the safety violation rate by more than two orders of magnitude compared to unaided or tremor-canceling baselines.
  • Accuracy: Recovered the 3× accuracy loss that occurred when only safety was enforced. While enforcing safety alone (AC-S) increased error to 1112 µm, adding guidance (AC-S+G) brought it back down to 380 µm—comparable to unaided performance but with near-perfect safety.
  • Efficiency: The optimization solve times were extremely low (median ~24 µs), easily fitting within a 1 kHz control loop.
Comparison of safety violation rates and tip RMSE across different conditions. AC-S+G maintains high safety without sacrificing accuracy.

Limitations and Future Work

  • Static Safety Margins: The current safety buffer is a fixed value. The authors suggest that an online, adaptive margin could better account for different users or levels of fatigue.
  • Pilot Study Size: The evaluation was limited to four novice subjects. Future work will involve trained experts and more complex, realistic surgical tasks.
  • Learning Effects: The authors noted that subjects' intuition for the "fulcrum effect" (where the tip moves opposite to the hand) varied, which might have confounded some of the performance data. Future studies will train users to "convergence" to eliminate this variable.

Summary updated

Abstract

In micromanipulation, erroneous hand motion limits unaided manipulation accuracy, which can affect task performance and safety in critical applications like microsurgery. Handheld robots (micromanipulators) enable motion accuracy beyond human limits by providing robotic assistance in a low-cost, compact, intuitive form factor. Particularly, active constraints are an effective form of robotic assistance, which aids the operator by restricting the robot's permissible motion based on task-specific requirements or objectives, thereby providing task guidance and enforcing safety. However, existing formulations of active constraints for handheld systems are limited, most notably because of the inability to provide guidance and enforce safety simultaneously. Implementing active constraints with optimization-based methods can address these limitations, but existing implementations are designed for free-standing robots, and neglect that handheld robots actuate relative to the operator's hand with a limited workspace. In this work, we propose an optimization-based active constraint framework for handheld micromanipulation that enables simultaneous guidance and safety enforcement. The formulation was implemented in a handheld robot and demonstrated on a medically relevant artificial task. Results show that the proposed formulation reduced safety violation rate by more than two orders of magnitude relative to baselines, while recovering the threefold accuracy loss from enforcing safety alone.
BibTeX
@misc{ho2026active,
  author = {Roger Ho and Robert A. MacLachlan and Cameron N. Riviere},
  title = {{Active Constraints for Simultaneous Guidance and Safety in Handheld Micromanipulation}},
  year = {2026},
  eprint = {2610.37981},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.37981},
}
Figure