cs.ROApr 9, 2026

EMMa: End-Effector Stability-Oriented Mobile Manipulation for Tracked Rescue Robots

Authors: Yifei Wang, Hao Zhang, Jidong Huang, Shuohang Fang, Haoyao Chen

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

The autonomous operation of tracked mobile manipulators in rescue missions requires not only ensuring the reachability and safety of robot motion but also maintaining stable end-effector manipulation under diverse task demands. However, existing studies have overlooked many end-effector motion properties at both the planning and control levels. This paper presents a motion generation framework for tracked mobile manipulators to achieve stable end-effector operation in complex rescue scenarios. The framework formulates a coordinated path optimization model that couples end-effector and mobile base states and designs compact cost/constraint representations to mitigate nonlinearities and reduce computational complexity. Furthermore, an isolated control scheme with feedforward compensation and feedback regulation is developed to enable coordinated path tracking for the robot. Extensive simulated and real-world experiments on rescue scenarios demonstrate that the proposed framework consistently outperforms SOTA methods across key metrics, including task success rate and end-effector motion stability, validating its effectiveness and robustness in complex mobile manipulation tasks.

Figures & tables

Fig. 1 : The concept of mobile manipulation for explosive ordnance disposal in rescue scenarios.
Fig. 2 : General framework of EMMa.
Fig. 3 : The typical mechanism of a tracked mobile manipulator and its key reference frame and state definition.
Fig. 4 : The typical arm span schematic description of a tracked mobile manipulator.
Fig. 5 : Manipulator configuration selection near obstacles. In the color-coded ESDF map, red squares indicate obstacles in the environment. The green arrow nk\mathbf{n}_{k} denotes the midline vector pointing toward the elbow, and the red arrow no\mathbf{n}_{o} represents the ESDF gradient direction at the midpoint of the line from end-effector to base.
Fig. 6 : Joint configuration and base path interpolation considering end-effector stability.
Fig. 7 : Four typical simulated rescue scenarios with distinct end-effector operation requirements: (a) obstacle-aware grasping with configuration adjustment; (b) grasping during locomotion; (c) environment inspection during locomotion; and (d) transportation of a force-sensitive payload over rough terrain.
ParametersDescriptionValue
ϵed,ϵbd,ϵad\epsilon_{ed},\epsilon_{bd},\epsilon_{ad}clearances of EE/Base/Arm0.05/0.2/0.1 m
ϵew\epsilon_{ew}arm workspace limitation0.85 m
δJ\delta_{J}Frobenius norm threshold3.0
Ωea,Ωeκ,Ωew\Omega_{ea},\Omega_{e\kappa},\Omega_{ew}weight of the EE costs102,102,10310^{2},10^{2},10^{3}
Ωbk\Omega_{bk}weight of the Base cost10210^{2}
Ωm\Omega_{m}weight of the Manipulability5050
TABLE I : Typical parameter settings
Fig. 8 : The typical performance of different algorithms in grasping near obstacles.
Fig. 9 : The typical end-effector executed trajectories of all compared methods across the four simulated tasks. The coordinate frames in (b-c) indicates the optimal grasp/inspection pose, respectively.
Fig. 10 : The typical executed trajectories of the end-effector in grasping during locomotion.
 Grasp during locomotionInspect during locomotion
 σla↓\sigma_{la}\downarrowσaa↓\sigma_{aa}\downarrowκmax⁡↓\kappa_{\max}\downarrowσla↓\sigma_{la}\downarrowσaa↓\sigma_{aa}\downarrowκmax⁡↓\kappa_{\max}\downarrow
Ours1.0901.658142.21.7261.50044.8
ReDyn1.3582.837349.41.5551.642133.2
GP1.4341.81488.62.9143.838177.6
TABLE II : Standard deviations of the end-effector linear acceleration σla(m/s2)\sigma_{la}\mathrm{(m/s^{2})} , angular acceleration σaa(rad/s2)\sigma_{aa}\mathrm{(rad/s^{2})} , and the maximum curvature κmax⁡(m−1)\kappa_{\max}(\mathrm{m^{-1}}) of the executed end-effector trajectories for the compared methods performing rescue tasks with end-effector fixed-pose constraints.
Fig. 11 : Raigor: a self-developed tracked mobile manipulator.
Fig. 12 : Real-world experiment setups.
Fig. 13 : The typical robot performance during the tasks: (a-b) the end-effector states in grasping during locomotion, captured either 0.1s0.1\mathrm{s} before gripper closure or at the moment when the end-effector is closest to the grasping pose; (c) the end-effector states near the objects during the object inspection task on rugged terrain.
Fig. 14 : The typical end-effector executed trajectories during the experiments. The coordinate frames in (a) and (b) indicate the optimal grasping poses.
 Easy SceneHard Scene
bvmaxx{}_{b}v_{max}^{x}≤0.2\leq 0.2≤0.3\leq 0.3≤0.5\leq 0.5≤0.2\leq 0.2≤0.3\leq 0.3≤0.5\leq 0.5
Ours100%100%100%100%60%40%
ReDyn100%80%0%0%0%0%
GP100%40%20%0%0%0%
TABLE III : Success rates of the different algorithms under varying chassis velocity limitations bvmaxx(m/s){}_{b}v_{max}^{x}(\mathrm{m/s}) and task difficulty setups.
Fig. 15 : The typical end-effector and payload motion traces in the ablation experiment ( 15s15\mathrm{s} in total).
Fig. 16 : The typical end-effector executed trajectories in the ablation experiment (top-down view).
Fig. 17 : The typical end-effector linear velocity and acceleration patterns in the ablation experiment.
 pmax↓p_{\mathrm{max}}\downarrowvmean↓v_{\mathrm{mean}}\downarrowvmax↓v_{\mathrm{max}}\downarrowamean↓a_{\mathrm{mean}}\downarrowamax↓a_{\mathrm{max}}\downarrow
F3B-P30.01690.00970.03260.05250.2382
F3B-P10.02360.01200.04070.05960.3041
FB-P30.05210.01470.06510.07470.5472
FB-P10.20520.04240.12100.09320.3130
TABLE IV : Quantitative performance of different control strategies in real-world experiments: pmax (m)p_{\mathrm{max}}\,(\mathrm{m}) denotes the Euclidean distance between the farthest waypoints along the executed end-effector trajectory, vmean (m/s)v_{\mathrm{mean}}\,(\mathrm{m/s}) denotes the mean linear velocity of the end-effector in the global coordinate frame, vmax (m/s)v_{\mathrm{max}}\,(\mathrm{m/s}) denotes the maximum linear velocity of the end-effector in the global coordinate frame, amean (m/s2)a_{\mathrm{mean}}\,(\mathrm{m/s^{2}}) denotes the mean linear acceleration of the end-effector in the global coordinate frame, and amax (m/s2)a_{\mathrm{max}}\,(\mathrm{m/s^{2}}) denotes the maximum linear acceleration of the end-effector in the global coordinate frame.
More in this topic

Mobile Manipulation

Mobile manipulation studies robots that navigate through environments while using arms or other effectors to interact with objects.

See all 91 papers

Explore similar work

CardsList
Apr 7, 2026

Synergizing Efficiency and Reliability for Continuous Mobile Manipulation

Humans seamlessly fuse anticipatory planning with immediate feedback to perform successive mobile manipulation tasks without stopping, achieving both high efficiency and reliability. Replicating this fluid and reliable behavior in robots remains fundamentally challenging, not only due to conflicts between long-horizon planning and real-time reactivity, but also because excessively pursuing efficiency undermines reliability in uncertain environments: it impairs stable perception and the potential for compensation, while also increasing the risk of unintended contact. In this work, we present a unified framework that synergizes efficiency and reliability for continuous mobile manipulation. It features a reliability-aware trajectory planner that embeds essential elements for reliable execution into spatiotemporal optimization, generating efficient and reliability-promising global trajectories. It is coupled with a phase-dependent switching controller that seamlessly transitions between global trajectory tracking for efficiency and task-error compensation for reliability. We also investigate a hierarchical initialization that facilitates online replanning despite the complexity of long-horizon planning problems. Real-world evaluations demonstrate that our approach enables efficient and reliable completion of successive tasks under uncertainty (e.g., dynamic disturbances, perception and control errors). Moreover, the framework generalizes to tasks with diverse end-effector constraints. Compared with state-of-the-art baselines, our method consistently achieves the highest efficiency while improving the task success rate by 26.67%--81.67%. Comprehensive ablation studies further validate the contribution of each component. The source code will be released.

Chengkai Wu, Ruilin Wang, Yixin Zeng +7

Robotics and Autonomous Systems Thrust, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China · School of Artificial Intelligence, Sun Yat-sen University, Zhuhai, China · Department of Mechanical and Energy Engineering, Southern University of Science and Technology, Shenzhen, China

Mobile ManipulationRobotic ControlMotion PlanningLong-Horizon Robotic ManipulationRobot Trajectory Tracking
AskPDF
Aug 7, 2026

A Reconfigurable Tracked Robot for Enhanced Obstacle Traversal Through Movable Articulation Point and Internal Mass Relocation

Tracked robots are widely used in unstructured environments; however, their obstacle traversal capability is fundamentally limited by a tradeoff between front-end reachability and locomotion stability. This study presents TRASER (Tracked Robot with Articulated Spine for Extended Reach), a reconfigurable tracked robot capable of relocating both its articulation point and internal mass. TRASER employs a tape-spring mechanism that localizes compliance to the bending region while maintaining high stiffness in the remaining body, thereby improving both front-end reachability and center-of-mass (CoM) shifting capability. Geometric and static models are developed to analyze the effects of articulation point and CoM position on step and ditch traversal performances. Experiments demonstrate step traversal, suspended-platform traversal, and ditch traversal of 74%, 66%, and 59% of the robot body length, respectively. To the best of our knowledge, these results represent the highest reported obstacle traversal capabilities among tracked mobile robots.

Yuki Uda, Yasutaka Nakashima, Motoji Yamamoto +1

Department of Mechanical Engineering, Kyushu University, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan · Graduate School of Informatics and Engineering, The University of Electro-Communications, 1-5-1 Chofugaoka, Chofu-shi, Tokyo, 182-8585, Japan

RoboticsTerrain-Aware Robot LocomotionRobot Locomotion
AskPDF
Oct 14, 2026

Hierarchical Residual Policy Learning for Real-World Mobile Manipulation with Sparse Human Guidance

Mobile Manipulation (MoMa) is a cornerstone of autonomous robotic systems, yet real-world fine-tuning remains a formidable challenge because of high-dimensional action spaces and intricate coupling between heterogeneous modules. In this paper, we propose HiRe-MoMa (Hierarchical Residual Mobile Manipulation), a systematic framework for stable and sample-efficient real-world fine-tuning of complex MoMa systems. At its core, HiRe-MoMa adopts a Hierarchical Residual Architecture that fine-tunes a frozen base policy. A Policy-Agnostic Residual learns additive corrections atop the base policy, while Hierarchical Decoupling splits the residual into mobile-base and manipulator-arm heads with unidirectional conditioning, mitigating gradient interference and aligning corrections with the platform's physical hierarchy. To improve sample efficiency in data-expensive real-world settings, we propose Expert-Guided Value Alignment, which leverages sparse human interventions with under 20% training coverage to reshape the critic's value landscape, effectively guiding autonomous exploration through expert priors. Extensive real-world experiments on the TIAGo++ platform demonstrate that HiRe-MoMa consistently enhances task performance with only around 30 minutes of online interaction, regardless of the base policy architecture. To our knowledge, this is among the first frameworks to enable efficient on-robot fine-tuning for mobile manipulation in the real world, providing a practical paradigm for deploying complex robotic systems in dynamic, human-centric environments. Further details can be found on our project website: https://hire-moma.github.io/

Zhefei Gong, Hefeifei Jiang, Zehao Wang +1

Northeastern University · University of California, Riverside · Georgia Institute of Technology

Mobile ManipulationRobotic RLResidual Policy LearningData-Efficient Robot LearningHuman-in-the-Loop AI
AskPDF