← Paper Digest

Toward Resilient Self-Driving Laboratories through Telerobotic Interventions

Ziheng Wu, Adrian Vecina Tercero, Andre A. Fatehi, Abhik Sarkar, Brian Giera, Eric B. Duoss, Christopher M. Spadaccini, Nikhil Deshpande, Aldair E. Gongora

November 2026

TL;DR A SpaceMouse interface lets a person take over a UR3e arm when a self-driving lab's automation fails. With six novice operators on three fault-recovery tasks, controlling pitch and yaw with separate keys raised success from 70.5% to 80.4% and cut task time by 25.6 s.

What they do

Self-driving labs run experiments with robot arms, but a dropped vial or a missing holder halts the workflow. This paper studies remote teleoperation as the recovery step: an operator takes over the lab’s UR3e with a 3D mouse and a wrist-camera feed. Two user studies test how the input mapping and reference-path guidance affect operators. The paper says it does not propose a new teleoperation algorithm.

How it works

  • Interface. SpaceMouse inputs map to end-effector Cartesian velocity, sent at 500 Hz over RTDE. Coupled mode puts all six DOFs on the central knob; decoupled mode keeps x, y, z and roll on the knob and moves pitch and yaw to four side keys.
  • Fault scenarios. Process interruption (finish a halted vial transfer), initialization error (fetch and place a missing vial holder) and escalated fault (grasp a dropped vial with the arm tilted), increasing in difficulty.
  • Risk analysis. Logged trials are scored afterwards with Yoshikawa manipulability, Jacobian condition number and minimum link distance, to locate singularity and self-collision risk.

Results

  • Controller study, 6 novices, 86 trials: decoupled mode raised success from 70.5% to 80.4% (process interruption 71% to 86%) and cut completion time by 25.6 s (p = 0.031). Its lower rotation was not significant (p = 0.094).
  • Guidance study, 8 operators, 12 trials each: unguided success was under 50% in the first two runs and 100% by run 6. The multi-waypoint reference path is 755 mm, twice the 375 mm direct path, yet operators imitating it moved only 373 mm in 38.5 s.
  • An operator restored a mock chemical-spill scene through teleoperation alone without causing new errors.

Limitations

Small, mostly novice participant pools and simplified faults. Guided trials always came after unguided ones, so practice is confounded with guidance. The risk metrics were computed offline, and there is no autonomous diagnosis, recovery or shared autonomy.

Figures

From Figure 1: (A) SpaceMouse key layout for the coupled and decoupled modes, (B) the hardware, with the operator wat...
From Figure 1: (A) SpaceMouse key layout for the coupled and decoupled modes, (B) the hardware, with the operator watching a wrist-camera feed and the UR3e on the lab side, and (C) the software linking them.
From Figure 2: Robot trajectories for the three simulated faults: (A) process interruption, a vial transfer halted mi...
From Figure 2: Robot trajectories for the three simulated faults: (A) process interruption, a vial transfer halted midway; (B) initialization error, a missing vial holder; (C) escalated fault, a dropped vial.
From Figure 3: Coupled vs. decoupled control on the three faults (PI, IE, EF): success rate, duration, path length an...
From Figure 3: Coupled vs. decoupled control on the three faults (PI, IE, EF): success rate, duration, path length and rotation. Error bars are one standard deviation; duration, path length and rotation use successful trials only.
From Figure 8: The
From Figure 8: The "telerobotic janitor" demo. Through teleoperation alone, the UR3e stacks a vial holder, moves flasks and a lattice, disposes of a sharp, cleans a spill, closes a trash bin and erases a whiteboard.
Abstract

Self-driving laboratories (SDLs) accelerate scientific research by integrating robotics, automated experimentation, and machine learning to execute experiments autonomously. Biology and chemistry have adopted these systems. However, increasing SDL complexity heightens susceptibility to unexpected faults in automated workflows, highlighting the need for efficient recovery mechanisms. In this study, we investigate telerobotic interventions to enhance SDL resilience against unexpected failures. We developed a teleoperation interface enabling human operators to resolve disruptions during automated tasks. The interface enables precise control of a six-degree-of-freedom (DOF) robotic arm (UR3e) using a 3D mouse (3Dconnexion SpaceMouse Pro). Two velocity-based controller strategies were evaluated: (i) a coupled mode mapping all six DOFs to the 3D mouse's central knob and (ii) a decoupled mode separating translational and rotational motions. To evaluate the framework, teleoperators intervened across three SDL fault scenarios. In the first study, comparative trials showed that the decoupled controller was associated with shorter completion times and reduced rotational motion. In a second study, teleoperators performed an aiming task with and without reference-path guidance. Users were more target-driven and risk-tolerant without guidance but showed behavioral differences under different guidance strategies. These findings suggest that controller configuration and reference-path design influence operator performance during SDL fault recovery, informing teleoperation interface optimization. This work demonstrates the potential feasibility of teleoperation as a human-in-the-loop recovery mechanism for SDL workflows, combining robotic precision with human adaptability and problem-solving. These findings also provide a foundation for future studies in shared autonomy and semi-autonomous intervention.

Summaries are written automatically from each paper's text. Figures and videos come from the paper and its project page; all credit goes to the authors.