IROS 2026 · Modular Robotics

MobiDock

Design and Control of a Modular Self-Reconfigurable
Bimanual Mobile Manipulator via Robotic Docking

Xuan Thuan Nguyen^†1Khac Nam Nguyen^†1Ngoc Duy Tran² Thi Thoa Mac¹Anh Nguyen³Hoang Hiep Ly¹Tung D. Ta^2,4

¹Hanoi University of Science and Technology · ²The University of Tokyo
³University of Liverpool · ⁴Keio University
^†Equal contribution

↗ Read the paper ⌘ Code ▶ Video { } BibTeX

Independent when flexibility matters. Physically unified when stability matters.

Video · MobiDock in action

From independent modules to one bimanual platform.

Watch the complete system demonstration, including autonomous alignment, physical docking, unified motion, and collaborative manipulation.

Watch on YouTube ↗

01 · Overview

A structural answer to multi-robot coordination.

Multi-robot mobile manipulators are flexible, but coordinating them during collaborative tasks introduces communication, control, and dynamic-stability challenges. MobiDock adds a physical docking mode: two independent robots autonomously connect and become one rigid bimanual platform.

A vision-guided approach and threaded screw-lock mechanism establish a zero-backlash connection. After docking, a unified control law treats the assembly as a single kinematic entity—reducing coordination overhead while preserving the option to separate for independent tasks.

Multi-modal hardware

Switch between independent mobile manipulators and a rigid, unified bimanual configuration.

End-to-end reconfiguration

Vision-based alignment, mechanical locking, and post-docking control in one framework.

Experimental evidence

Benchmarking shows improved stability, angular precision, and task efficiency.

02 · System

Align. Connect. Move as one.

Each module combines a 6-DOF arm with a triangular three-wheel omnidirectional base. The base-mounted docking interface enables rigid physical reconfiguration without sacrificing mobility.

Exploded view of the screw-based docking mechanism — Mechanical designA rotating male hub engages a stationary female hub to create a rigid threaded connection.

MobiDock autonomous docking procedure — Docking lifecycleSearch, coarse alignment, visual servoing, contact, and locking.

AprilTag and camera coordinate axes — Vision guidanceAprilTag offsets guide lateral centering and depth control during approach.

1DetectLocate the target AprilTag

2AlignCorrect lateral and angular offsets

3ApproachClose the remaining depth

4LockEngage the screw mechanism

Four-module MobiDock configurations — The generalized control model supports multiple docking geometries by transforming each module frame to the reconfigured system frame.

03 · Experiments

Rigid coupling, measurable gains.

Four experiments evaluate autonomous docking, post-reconfiguration motion, dynamic stability, and collaborative task performance.

93.3%Docking success

15 trials across four lighting conditions

63.3%Lower RMS acceleration

0.2226 vs. 0.6068 m/s²

49%Better angular precision

0.56° vs. 1.10° deviation

76sAverage transport time

Compared with 84 s cooperating

Docking trials under different lighting conditions — **Robust autonomous docking.** Reliable tag detection across common indoor lighting, with a ±60° viewing threshold.

Path tracking comparison for single and docked robots — **Single-platform behavior.** The docked six-wheel platform maintains tracking performance comparable to one module.

Dynamic stability comparison results — **Smoother payload transport.** Ten trials show lower RMS acceleration, jerk, and angular deviation.

Trash collection task and completion-time results — **Faster collaborative work.** Docking removes continuous base-to-base synchronization during a dual-arm task.

MetricCooperatingDocked

RMS acceleration0.6068 m/s²0.2226 m/s²

Jerk2.038 m/s³1.710 m/s³

Angular deviation1.10°0.56°

Average time84 s76 s

04 · Citation

Build on MobiDock.

If you find this work useful, please cite the paper.

@inproceedings{nguyen2026mobidock,
  title     = {MobiDock: Design and Control of a Modular
               Self-Reconfigurable Bimanual Mobile Manipulator
               via Robotic Docking},
  author    = {Nguyen, Xuan Thuan and Nguyen, Khac Nam and
               Tran, Ngoc Duy and Mac, Thi Thoa and Nguyen, Anh and
               Ly, Hoang Hiep and Ta, Tung D.},
  booktitle = {IEEE/RSJ International Conference on
               Intelligent Robots and Systems (IROS)},
  year      = {2026}
}