MorphoGuard: A Morphology-Based Whole-Body Interactive Motion Controller
Authors: Chenjin Wang, Zheng Yan, Yanmin Zhou, Runjie Shen, Bin He
Organizations: Shanghai Research Institute for Intelligent Autonomous Systems Shanghai 201210, China · College of Electronics and Information Engineering (Tongji University) Shanghai 201210, China
Whole-body control (WBC) has demonstrated significant advantages in complex interactive movements of high-dimensional robotic systems. However, when a robot is required to handle dynamic multi-contact combinations along a single kinematic chain-such as pushing open a door with its elbow while grasping an object-it faces major obstacles in terms of complex contact representation and joint configuration coupling. To address this, we propose a new control approach that explicitly manages arbitrary contact combinations, aiming to endow robots with whole-body interactive capabilities. We develop a morphology-constrained WBC network (MorphoGuard)-which is trained on a self-constructed dual-arm physical and simulation platform. A series of model recommendation experiments are designed to systematically investigate the impact of backbone architecture, fusion strategy, and model scale on network performance. To evaluate the control performance, we adopt a multi-object interaction task as the benchmark, requiring the model to simultaneously manipulate multiple target objects to specified positions. Experimental results show that the proposed method achieves a contact point management error of approximately 1 cm, demonstrating its effectiveness in whole-body interactive control.
Figures & tables
Figure 1: Schematic of morphology-based whole-body motion control (MorphoGuard). (A) An example of a robot facing complex contact combinations in an unstructured environment. (B) MorphoGuard is tasked with learning a mapping from the entity morphology in the workspace to the configuration space. (C) MorphoGuard takes as input MPs that discretely represent the robot morphology and predicts the corresponding joint motion commands.
Figure 2: An overview of the MorphoGuard framework. The model consists of a MPs encoding module, an fusion module, and a joint command decoding module. The morphology encoding module encodes the current morphology and the target morphology into feature representations. The information fusion module integrates the encoded features to capture the relationship between the current and target morphologies. Finally, the joint command decoding module predicts the joint commands based on the fused features.
Figure 3: Simulation and physical platform used for data collection.
Figure 4: Training and validation loss curves for different backbone architectures. The MLP backbone demonstrates the lowest validation loss, indicating superior generalization performance compared to CNN, Transformer, and GNN backbones.
Model Name
Params (M)
Backbone Config
FLOPs (G)
Inference Time (ms)
(Layers × Width)
(batch=64)
small_5m
5.9
8 × 512
0.75
0.8–1.2
medium_10m
10.2
8 × 768
1.95
1.2–1.5
medium_25m
25.9
10 × 1024
3.95
4.5–6.5
large_50m
51.9
22 × 1024
6.64
7.5–10.5
large_100m
115.4
22 × 1536
15.05
18–25
Table 2: MLP Scaling Experiment: Model Architecture and Computational Efficiency
Model Size
Val Loss
Val Std
Rel Improvement (%)
p-value
Sig
Scale Small 5M
0.0292
0.0229
0.00
nan
ns
Scale Medium 10M
0.0178
0.0208
39.02
0.0204
Scale Medium 25M
0.0116
0.0151
60.27
0.0204
Scale Large 50M
0.0096
0.0127
67.18
0.0257
Scale Large 100M
0.0095
0.0120
67.50
0.0291
Scale Xlarge 200M
0.0104
0.0133
64.36
0.0234
Table 3: Model Scale Comparison Results
Fusion Method
Val Loss
Val Std
Rel Improvement (%)
p-value
Sig
Additive
0.0159
0.0211
0.00
—
—
Concat
0.0232
0.0248
-45.81
0.1112
Input Concat
0.0171
0.0230
-7.86
0.2473
Parallel Branch
0.0321
0.0218
-102.05
0.0407
∗
Adaptive Norm
0.0178
0.0229
-11.90
0.1015
Table 4: Fusion Method Comparison Results
Figure 5:
Figure 6: Test procedure and object trajectories for the sequential manipulation task. (A)sequential manipulation task. (B)simultaneous manipulation task.
Figure 7: Test results for the multi-object manipulation tasks. (A) Contact point error for the sequential manipulation task. (B) Contact point error for the simultaneous manipulation task.
Cooperative object transport in unstructured environments remains challenging for assistive humanoids because strong, time-varying interaction forces can make tracking-centric whole-body control unreliable, especially in close-contact support tasks. This paper proposes a bio-inspired, interaction-oriented whole-body control (IO-WBC) that functions as an artificial cerebellum - an adaptive motor agent that translates upstream (skill-level) commands into stable, physically consistent whole-body behavior under contact. This work structurally separates upper-body interaction execution from lower-body support control, enabling the robot to maintain balance while shaping force exchange in a tightly coupled robot-object system. A trajectory-optimized reference generator (RG) provides a kinematic prior, while a reinforcement learning (RL) policy governs body responses under heavy-load interactions and disturbances. The policy is trained in simulation with randomized payload mass/inertia and external perturbations, and deployed via asymmetric teacher-student distillation so that the student relies only on proprioceptive histories at runtime. Extensive experiments demonstrate that IO-WBC maintains stable whole-body behavior and physical interaction even when precise velocity tracking becomes infeasible, enabling compliant object transport across a wide range of scenarios.
Hao Zhang, Yves Tseng, Ding Zhao +1
Department of Electrical Engineering, the University of Texas at Arlington, 76010 Arlington, USA · Department of Mechanical Engineering, Carnegie Mellon University, 15213 Pittsburgh, USA
Current humanoid dexterous manipulation remains largely confined to end-effectors and simple grasps, falling short of humans' ability to engage the entire body by bracing and sliding bulky objects against the arms and torso. We introduce Whole-Body Dexterity (WBD), a general framework that uses the robot's body and appendages as contact surfaces to grasp and manipulate objects. Our sim-to-real pipeline first generates a dataset of feasible whole-body grasps, then trains reinforcement-learning experts on a general any-to-any repositioning objective by sampling initial and goal grasps from the dataset, and distills these experts into a single policy that uses proprioception and object geometry. We show that the distilled policy transfers zero-shot to a real Unitree G1, where it repositions a basketball and a cardboard box to regions around the torso and completes a long-horizon task that stows a box under one arm, carries it from a shelf to a table, and places it. The controller operates interactively, steered by coarse motion references from a human teleoperator. Across real-world whole-body manipulation tasks, the distilled controller completes 90% of trials, compared with 18% for direct retargeting and 4% for a generalist whole-body controller.
Conor Mc Gartoll, An Dang, Pranay Thangeda +5
Amazon · University of California, San Diego · University of Michigan
Achieving versatile and natural whole-body humanoid interaction control remains challenging due to the high cost of whole-body teleoperation data. We present ZeroWBC, a teleoperation-free framework that learns humanoid whole-body interaction from human egocentric videos paired with synchronized whole-body motion and text annotations. ZeroWBC adopts a generation-then-tracking formulation to tackle the static scene whole-body interaction control problem. Given an initial egocentric image and a language instruction, a fine-tuned Vision-Language Model generates future human whole-body motion tokens, which are decoded into continuous motions and retargeted to the humanoid. The resulting reference motions, together with root and key body-part trajectories, are then executed by a general interactive motion tracking policy. To improve interaction performance, we introduce an interaction-oriented tracking reward that prioritizes global root and key body-part trajectory alignment while preserving natural whole-body motion. Experiments on the Unitree G1 humanoid robot show that ZeroWBC enables diverse scene-aware behaviors without robot teleoperation demonstrations. These results suggest a scalable paradigm for learning natural humanoid whole-body interaction from human egocentric data.
Haoran Yang, Jiacheng Bao, Yucheng Xin +5
University of Science and Technology of China · Shanghai AI Laboratory · Tsinghua University +2