Organizations: School of Artificial Intelligence and Robotics, Hunan University, China · National Engineering Research Center of Robot Visual Perception and Control Technology, Hunan University, China · College of Semiconductors (College of Integrated Circuits), Hunan University, Changsha, China
Quadruped mobile manipulation requires two tightly coupled capabilities: reaching manipulation-ready configurations and maintaining stable contact throughout articulated-object interaction. However, existing methods often terminate navigation near the target, leaving a gap between reachability and manipulation readiness, while tracking lag, motion jitter, and contact instability limit continuous interaction. To address these challenges, we present TONAV, a unified framework integrating task-oriented navigation with action-velocity chunk learning. First, we introduce a position-velocity-coupled teleoperation framework that explicitly captures motion dynamics to improve master-follower consistency and collect smooth, temporally consistent demonstrations. Next, task-oriented navigation leverages vision-language reasoning to decompose high-level instructions into executable subgoals and adaptively refine the robot base toward a manipulation-ready configuration. Finally, action-velocity chunk learning jointly models joint positions and their temporal transitions under velocity supervision, enabling smooth and stable sustained-contact manipulation. Real-world experiments across diverse articulated-object tasks demonstrate that TONAV achieves higher success rates in both task-oriented navigation and complete mobile manipulation, mitigating the navigation-manipulation gap and improving continuous-contact interaction. The project page is at https://haochen611.github.io/TONAV.
Figures & tables
Fig. 1 : Illustration of articulated-object quadrupedal mobile manipulation, with progressive target approaching followed by position-velocity-based sustained-contact manipulation for smoother and more stable interaction.
Fig. 2 : Comparison of position-only and position-velocity-coupled teleoperation. (a) Teleoperation at increasing speeds, showing reduced tracking lag with position-velocity coupling. (b) Master-follower trajectories at 1× and 2× speeds, showing lower latency and better temporal consistency.
Fig. 3 : Position-velocity-coupled teleoperation with 220Hz position-velocity tracking, 30Hz camera observations, and timestamp-based alignment across latency-varying modalities.
Fig. 4 : Overview of TONAV with an upper navigation branch and a lower manipulation branch. The upper branch combines hierarchical task reasoning and adaptive last-mile navigation to reach a manipulation-ready configuration, while the lower branch learns action-velocity chunks from motion-consistent demonstrations for smooth and stable sustained-contact manipulation.
Fig. 5 : Illustration of the Adaptive Approach Module (AAM). AAM performs coarse approaching with a diffusion policy and fine target refinement with visual feedback.
Fig. 6 : Qualitative comparison on the two tasks. Ours leverages action–velocity demonstrations and chunk learning to achieve smoother motion and more stable sustained-contact manipulation than ACT [ 28 ] and DP [ 29 ] .
Navigation Success Rates
Overall
Method
Close Drawer
Turn On Lamp
Lower Toilet Lid
SR
StreamVLN [ 7 ]
0/5
1/5
1/5
13.33%
InternNav [ 43 ]
2/5
1/5
1/5
26.67%
TONAV (Ours)
4/5
3/5
2/5
60.00%
TABLE I : Comparison of navigation Success Rates (SR) in the first stage across three articulated-object loco-manipulation tasks, together with the Overall SR.
Fig. 7 : Qualitative comparison of task-oriented navigation. Ours progressively approaches the target, refines the base pose for manipulation readiness, and closes the drawer within the manipulable workspace using AVTF-based control (see Sec. III-A ).
Fig. 8 : Joint-angle trajectories of Piper Joints 1–3 during the Close Drawer task. Ours yields smoother trajectories with reduced oscillation than ACT [ 28 ] and DP [ 29 ] .
Fig. 9 : Real-world demonstration of AAM. The first two frames show coarse approach, and the last two refine base orientation and target distance for manipulation readiness.
Task Success Rates
Overall
Method
Close Drawer
Turn On Lamp
Lower Toilet Lid
SR
ACT [ 28 ] (w/o P-V Control)
1/5
0/5
1/5
13.33%
DP [ 29 ] (w/o P-V Control)
1/5
1/5
0/5
13.33%
TONAV (w/o P-V Control)
2/5
2/5
4/5
53.33%
ACT [ 28 ]
2/5
3/5
2/5
46.67%
DP [ 29 ]
2/5
1/5
1/5
26.67%
TABLE II : Comparison of complete mobile manipulation success rates, with all methods sharing a single navigation run to the manipulation-ready region. ( w/o P-V Control : demonstrations collected without position-velocity-coupled teleoperation.)
Instruction Decomposition Success Rates
Overall
Method
Close Drawer
Turn On Lamp
Lower Toilet Lid
SR
Doubao-Seed-2.1-Pro (w/o PP-CoT)
1/5
1/5
0/5
13.33%
Qwen-3.7-Max (w/o PP-CoT)
2/5
1/5
1/5
26.67%
Doubao-Seed-2.1-Pro
3/5
2/5
2/5
46.67%
Qwen-3.7-Max (Ours)
4/5
3/5
2/5
60%
TABLE III : Ablation of PP-CoT and comparison of different LLMs for manipulation-oriented navigation across three articulated-object mobile manipulation tasks.
Mobile manipulation requires a robot to navigate to a target object or receptacle and then perform intended manipulation. However, reaching the vicinity of the target does not guarantee a manipulation-ready base pose, a problem known as last-mile navigation. Prior methods for last-mile navigation either rely on manual pose annotation or task-specific training, limiting their scalability to open-vocabulary settings with fine-grained spatial constraints. We propose UniLM-Nav, a unified framework for zero-shot open-vocabulary last-mile navigation. UniLM-Nav decomposes last-mile navigation into view selection, task-conditioned affordance grounding, and geometry-aware base-pose reasoning, all resolved with a shared multimodal large language model (MLLM) backend. Specifically, UniLM-Nav first selects a reference view that best captures the target object or receptacle from recently collected observations. It then grounds task-relevant affordance point in the selected view and lifts the result into the robot-centric coordinate frame. Finally, conditioned on the grounded affordance, task context, and robot geometry, it infers a manipulation-ready base pose for the robot. We evaluate UniLM-Nav on the OVMM benchmark, where it outperforms the previous state-of-the-art method, MoTo, by 3.13 percentage points. Analyses show that the components of our method are crucial to final performance, and that the choice of MLLM also has a substantial effect. We further deploy UniLM-Nav on a Unitree B2 quadruped robot with a 6-DoF Unitree Z1 manipulator, validating its applicability to real-world mobile manipulation tasks.
Zhuofan Zhang, Tianxu Wang, Guoxi Zhang +6
1Tsinghua University · 2State Key Laboratory of General Artificial Intelligence, BIGAI · 3Harbin Institute of Technology +1
Quadrupedal loco-manipulation is commonly built on visual perception and proprioception. Yet reliable contact-rich manipulation remains difficult: vision and proprioception alone cannot resolve uncertain, evolving interactions with the environment. Tactile sensing offers direct contact observability, but scalable tactile-aware learning framework for quadrupedal loco-manipulation is still underexplored. In this paper, we present a tactile-aware loco-manipulation policy learning pipeline with a hierarchical structure. Our approach has two key components. First, we leverage real-world human demonstrations to train a tactile-conditioned visuotactile high-level policy. This policy predicts not only end-effector trajectories for manipulation, but also the evolving tactile interaction cues that characterize how contact should develop over time. Second, we perform large-scale reinforcement learning in simulation to learn a tactile-aware whole-body control policy that tracks diverse commanded trajectories and tactile interaction cues, and transfers zero-shot to the real world. Together, these components enable coordinated locomotion and manipulation under contact-rich scenarios. We evaluate the system on real-world contact-rich tasks, including in-hand reorientation with insertion, valve tightening, and delicate object manipulation. Compared to vision-only and visuotactile baselines, our method improves performance by 28.54% on average across these tasks.
Pokuang Zhou, Yuhao Zhou, Quan Khanh Luu +7
Purdue University, West Lafayette, IN, USA. · Istituto Italiano di Tecnologia, Genoa, Italy. · Columbia University, New York, NY, USA.
Achieving precise positioning of the mobile manipulator's base is essential for successful manipulation actions that follow. Most of the RGB-based navigation systems only guarantee coarse, meter-level accuracy, making them less suitable for the precise positioning phase of mobile manipulation. This gap prevents manipulation policies from operating within the distribution of their training demonstrations, resulting in frequent execution failures. We address this gap by introducing an object-centric imitation learning framework for last-meter navigation, enabling a quadruped mobile manipulator robot to achieve manipulation-ready positioning using only RGB observations from its onboard cameras. Our method conditions the navigation policy on three inputs: goal images, multi-view RGB observations from the onboard cameras, and a text prompt specifying the target object. A language-driven segmentation module and a spatial score-matrix decoder then supply explicit object grounding and relative pose reasoning. Using real-world data from a single object instance within a category, the system generalizes to unseen object instances across diverse environments with challenging lighting and background conditions. To comprehensively evaluate this, we introduce two metrics: an edge-alignment metric, which uses ground truth orientation, and an object-alignment metric, which evaluates how well the robot visually faces the target. Under these metrics, our policy achieves 74.58% success in edge-alignment and 89.42% success in object-alignment when positioning relative to unseen target objects. These results show that precise last-meter navigation can be achieved at a category-level without depth, LiDAR, or map priors, enabling a scalable pathway toward unified mobile manipulation. Project page: https://rpm-lab-umn.github.io/category-level-last-meter-nav/