Organizations: Faculty of Information Technology and Electrical Engineering, University of Oulu, 90570 Oulu, Finland · Research Unit of Health Sciences and Technology (HST), Faculty of Medicine, University of Oulu, 90220 Oulu, Finland · Research Unit of Disease Network, Faculty of Biochemistry and Molecular Medicine, University of Oulu, 90220 Oulu, Finland · School of Information Science and Engineering, Provincial Key Laboratory of Informational Service for Rural Area of Southwestern Hunan, Shaoyang University, 422000 Shaoyang, China · College of Computer Science and Engineering, Jishou University, 416000 Jishou, China · Department of Computer Science, University of Exeter, Exeter, United Kingdom · Department of Information Technology and Electrical Engineering, University of Naples Federico II, Naples, Italy · VTT Technical Research Centre of Finland, 90570 Oulu, Finland
Robotic haircutting requires controlled tool motion near the head while simultaneously accounting for communication, visual feedback, tool actuation, and interruption handling. Existing studies still lack an operator-in-the-loop reference for analyzing these coupled behaviors before human trials or stronger autonomy. This paper presents TeleHairing, a closed-loop teleoperation architecture for mannequin-based robotic haircutting evaluation under local, relay, and remote deployment conditions. Logged timing shows that the main remote latency increase occurs before the robot-side control endpoint: overall timing reached 190.5~ms in remote mode, while robot-side command queue, control processing, and control-to-robot timing remained similar across modes. Trajectory analysis shows that the larger remote command-following error was dominated by the terminal withdrawal segment rather than accumulated uniformly over the path; excluding this segment reduced remote root-mean-square error (RMSE) from 45.1 mm to 9.6 mm. Detection-loss trials further show that rebase events resumed motion without a large target jump under the tested condition. These results clarify how deployment, execution, and interruption affect the robotic haircutting teleoperation loop, providing a quantitative reference for future autonomy, safety, and user-facing studies.
Figures & tables
Figure 1: Architecture overview of TeleHairing. On the operator side, a phone camera supports pose estimation, and teleoperation messages are routed either directly or through a cloud relay to the robot-side control server. The control server coordinates manipulator and tool commands, while a feedback camera streams video back to the browser client to close the teleoperation loop. Orange numbered markers denote send-side or stage-start timestamps, whereas purple numbered markers denote receive-side or stage-end timestamps. The numbered pairs identify the timestamped timing segments reported in Table 2 .
Figure 2: Robot-side clipper integration used in TeleHairing.
Item
Configuration
Robot platform
UR10e 6-degree-of-freedom manipulator
End effector
Commercial hair clipper mounted through a dedicated 3D-printed adapter and electrically switched through a relay and STM32 interface
Control command interface
RTDE-based Cartesian command streaming
Controller frequency
30 Hz robot-side Cartesian command updates
Velocity limits
0.05 m/s translational limit and 15 deg/s rotational limit
Visual reference
Single AprilTag reference marker: tag36h11, ID 0, physical tag size 75 mm
Table 1: System and experimental configuration used in the TeleHairing evaluation.
Timing segment
Timestamp pair
Local mode
Relay mode
Remote mode
Browser → master
1–2
7.4 ± 2.2
9.3 ± 4.0
1.3 ± 3.8
Master proc
2–3
0.1 ± 0.3
0.1 ± 0.6
0.2 ± 0.4
Master → cloud
3–4
—
14.4 ± 2.7
116.9 ± 2.0
Cloud proc
4–5
—
0.0 ± 0.2
0.0 ± 0.2
Cloud → control
5–6
—
5.5 ± 2.2
5.7 ± 4.0
Master → control
3–6
1.1 ± 0.8
—
—
Table 2: Stage-wise communication timing across the three TeleHairing modes. Values are reported in milliseconds.
Figure 3: Representative command and measured trajectory overlays for the three TeleHairing modes.
Mode
N
RMSE (mm)
Peak error (mm)
Local mode
10
6.7 ± 1.7
27.3 ± 12.9
Relay mode
10
5.1 ± 0.9
19.6 ± 6.5
Remote mode
10
45.1 ± 17.1
223.6 ± 80.8
Remote mode (first 90%)
10
9.6 ± 3.0
34.9 ± 10.5
Table 3: Command-following trajectory error across the three TeleHairing modes.
Figure 4: Detection-loss recovery behavior in the dedicated trials. Panel (a) shows a single recovery event with the tracking gap aligned to the rebase instant; panel (b) shows the post-recovery median and interquartile displacement profile; and panel (c) shows the relation between tracking-gap duration, accumulated hold drift, and the first resumed command step.
Indicator
Value
Recorded events
23 events from 5 trials
Effective tracking gap
1.97 ± 1.25 s; range 0.43–5.47 s
Missing updates
53.61 ± 34.71
Measured hold drift
6.47 ± 1.50 mm
First resumed command step
0.64 ± 0.14 mm
Table 4: Quantitative indicators for detection-loss recovery.
Department of Mechanical and Aerospace Engineering, University of California at Irvine, Irvine, CA 92697 USA · Department of Electrical Engineering and Computer Science, University of California at Irvine, Irvine, CA 92697 USA · Professional Master of Embedded and Cyber-physical Systems Program, University of California at Irvine, Irvine, CA 92697 USA