Point It, Strike It: Direction-Conditioned Dynamic Manipulation of Deformable Linear Objects

Yi Yang; Xiang Fei; Lehong Wang; Zilin Dai; Ruogu Li; Jiting Cai; Liyao Chang; Xinyi Yang

cs.ROOct 7, 2026

Point It, Strike It: Direction-Conditioned Dynamic Manipulation of Deformable Linear Objects

Authors: Yi Yang, Xiang Fei, Lehong Wang, Zilin Dai, Ruogu Li, Jiting Cai, Liyao Chang, Xinyi Yang, +4 more

Organizations: The Robotics Institute, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, PA 15213 USA. · School of Ocean and Civil Engineering, Shanghai Jiao Tong University, 800 Dongchuan Rd, Minhang District, Shanghai, China. · Zhiyuan College, Shanghai Jiao Tong University, 800 Dongchuan Rd, Minhang District, Shanghai, China. · John A. Paulson School of Engineering and Applied Sciences, Harvard University, 29 Oxford Street, Cambridge, MA 02138 USA.

Abstract

Goal-conditioned dynamic manipulation of deformable linear objects has mainly specified goals as positions for a rope tip to reach. Many tasks, however, depend on how the tip arrives. We therefore study single-swing rope striking with goals that specify the tip's 3D position and arrival direction, across the workspace and on different ropes. This is challenging because rope dynamics are hard to model, no demonstrations exist, distinct swings reach the same goal with different reliability, and the sim-to-real gap extends beyond the rope. To address these challenges, we extend the state-of-the-art DLO simulator DeformX with GPU acceleration, a stable Cosserat rod solver, and a cross-flow aerodynamic model, yielding DeformX2.0, which is more than $20{,}000\times$ faster. We then propose TRACE (Trace-rooted Adaptive Cross-Entropy), which generates striking data by warm-starting each new target from the stored swing whose tip path passes closest to it. Its cost penalizes rope bending and abrupt tip motion to favor repeatable swings. A conditional flow-matching policy trained on this data reaches 92.1% accuracy in simulation. Finally, we propose RECAP (Residual Calibration Policy), which fits the simulator's rope and rig parameters to a few calibration swings and adapts actions with a correction policy trained in simulation. On a real robot, across three ropes, RECAP raises success within 5cm from 72% to 87% for position goals, and within 10cm and 10° from 50% to 79% for goals that also specify the arrival direction.

Figures & tables

Fig. 2: System overview and network architectures. Left: TRACE generates goal–action pairs in GPU co-simulation to train the base policy. RECAP calibrates the real system once and corrects the nominal action for each goal before a single open-loop strike. (a) The base and correction policies use conditional flow matching (CFM), conditioned on $g$ and $(g,\bar{w},\Delta,\hat{\eta})$ , respectively. (b) RECAP iteratively fits rope and rig parameters to measured calibration swings. These estimates condition the correction policy to adapt actions to the real system. Dashed lines indicate input.

Fig. 3: Experimental setup and simulation results. (a) Targets are selected within the workspace. (b) Arrival angle $\theta$ , measured after projecting the rope-tip velocity onto the tangent plane of the arm-centered sphere. (c) Radar-plot radius indicates data-generation coverage for each $\theta$ , and color indicates policy hit rate. (d,e) Data-generation coverage and policy accuracy, measured by solved generation targets and hit evaluation targets, respectively.

		3D goal (position)					4D goal (position and direction)
Rope	Method	SR@5	SR@10	Distance	Best	Worst	SR@5	SR@10	Distance	Best	Worst	Angle
		(%)	(%)	(cm)	(cm)	(cm)	(%)	(%)	(cm)	(cm)	(cm)	( ∘ )
Rope A	Base	68	91	$4.8_{\pm 4.9}$	$1.3_{\pm 1.1}$	$15.4_{\pm 5.9}$	23	41	$13.4_{\pm 12.0}$	$1.4_{\pm 0.5}$	$38.7_{\pm 25.4}$	$9.8_{\pm 17.3}$
	Baseline	65	96	$4.4_{\pm 2.8}$	$1.3_{\pm 0.4}$	$11.9_{\pm 2.5}$	31	45	$11.7_{\pm 9.3}$	$2.6_{\pm 0.4}$	$31.0_{\pm 0.7}$	$9.4_{\pm 13.8}$
	Ours	89	99	$\mathbf{2.8_{\pm 2.6}}$	$\mathbf{1.0_{\pm 0.8}}$	$\mathbf{9.8_{\pm 7.0}}$	61	75	$\mathbf{5.2_{\pm 4.3}}$	$\mathbf{0.7_{\pm 0.1}}$	$\mathbf{15.2_{\pm 2.3}}$	$\mathbf{4.5_{\pm 5.2}}$
Rope B	Base	83	97	$3.4_{\pm 2.8}$	$0.9_{\pm 0.3}$	$10.4_{\pm 1.7}$	33	55	$9.5_{\pm 7.5}$	$1.4_{\pm 0.7}$	$26.1_{\pm 2.2}$	$8.3_{\pm 11.9}$

TABLE II: Hardware striking on three ropes, 25 targets

\times

3 swings per cell. Base: nominal simulator; Baseline: simulator identified by a learned network as in Wiggle&Go [ 6 ] , no correction; Ours: RECAP. SR@

d

: success rate (%) within

d

of the target; 4D goals also require arriving within

10^{\circ}

of the commanded direction at

\geq 1

m/s along it. Distance and Angle (arrival-angle error): mean over all swings, with the std as a subscript. Best and Worst: the lowest and the highest per-target mean distance, with that target’s std. Mean is the mean over the three ropes. Best method per rope and goal type in bold.

Engine	Envs	Update rate (Hz)	Step time (ms)	Sim-to-wall time ratio
DeformX [ 2 ] (CPU, 1 thread)	1	12.9	77.5	0.22
Ours (RTX 4090)	1	136	7.4	2.3
Ours (RTX 4090)	1024	104	9.6	1780
Ours (RTX 4090)	2048	77.3	12.9	2640
Ours (RTX 4090)	8192	34.5	29.0	4710

TABLE I: Simulation throughput on the same 1 m, 20-segment rope, on one RTX 4090. Rates are per control step of the whole batch.

Fig. 4: Workspace maps on Rope A (real world) for 3D goals with Base, Baseline and Ours, and for 4D goals with Baseline and Ours: success rate (upper two rows) and mean tip–target distance (lower two rows) of each target over its three swings. A 3D goal succeeds within $5$ cm; a 4D goal, which also requires arriving within $10^{\circ}$ of the commanded direction at $\geq 1$ m/s, is scored within $10$ cm, as the column titles say. The distance scale saturates at $20$ cm. Each quantity is shown in the front view (lateral $x$ vs. height $z$ from the wall-mounted base) and, directly below it on the same $x$ axis, in the top view (lateral $x$ vs. forward $y$ ). Dots are the evaluated targets; the shaded field is a kernel-weighted interpolation between them, and the black star marks the wall-mounted robot base.

Method	Params	One-shot	Best-of-64	Miss	Angle
	(M)	(%)	(%)	(cm)	( ∘ )
Ours	11.32	73.5	92.1	3.6	2.4
Naive FM	11.35	67.9	88.9	5.3	3.0
Regression	11.31	21.4	–	24.0	12.4
Nearest-neighbor	—	33.6	–	9.2	6.4

TABLE III: Learning-method ablation : We compare our model, naive flow matching, and our network with a regression head at matched capacity, alongside nearest-neighbor lookup in the training data. Miss and angle are means over the one-shot sample on

n=5700

held-out goals. Regression falls below the lookup because it averages incompatible motion patterns into invalid swings.

Fig. 5: Learning multiple swings for one goal. We compare goal conditioning in every block (ours, left) with conditioning only at the input (naive FM, right). Stars mark two training swings $w_{A},w_{B}$ associated with goal $g$ ; dots show generated samples. Shading shows a 2D slice of the conditional log-density $\log p(w\mid g)$ , and arrows show the learned flow field at $s=0.9$ . (a,b) For swings from the same motion pattern, our model produces more concentrated samples. (c,d) For swings from different patterns, our model produces more clearly separated density peaks. Regression ( $\times$ ) predicts a single action that can average incompatible swings. (e) For the pair in (c,d), curves show the fraction of samples within a given distance of the nearer training swing, normalized by the pair’s separation in normalized action space.

Generator	Coverage (%) $\uparrow$	One-shot (%) $\uparrow$	Best-of-64 (%) $\uparrow$	Mean $L_{\mathrm{curv}}$ $\downarrow$
CEM w/ example	49.5	58.5	68.5	0.40
TRACE w/o example	96.8	75.5	83.9	0.51
TRACE	98.3	73.5	92.1	0.33

TABLE IV: Data-generation ablation evaluating workspace coverage, policy accuracy on 5700 random test goals, and trajectory quality measured by mean curvature loss. One-shot evaluates one policy sample; best-of-64 selects the candidate with the lowest TRACE loss. The example seeds higher-quality motion patterns, while TRACE improves coverage and accuracy of the dataset.

Explore similar work

RopeFormer: Cross-Trial Adaptation from Interaction History for Dynamic Rope Manipulation
Sep 20, 2026Menglin Wu, Kaixiang Yao, Shangbo Luan +2Dynamic Rope Manipulation Interaction History
University of California, Berkeley · Xi’an Jiaotong University · Southern University of Science and Technology +1
DeformX: A Versatile Co-Simulation Framework for Deformable Linear Objects
Jun 20, 2026Yi Yang, Xiang Fei, Lehong Wang +5Deformable Linear Objects Deformable
The Robotics Institute, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, PA 15213 USA. · School of Ocean and Civil Engineering, Shanghai Jiao Tong University, 800 Dongchuan Rd, Minhang District, Shanghai, China. · Zhiyuan College, Shanghai Jiao Tong University, 800 Dongchuan Rd, Minhang District, Shanghai, China. +2
DLO-Lab: Benchmarking Deformable Linear Object Manipulations with Differentiable Physics
Jun 2, 2026Junyi Cao, Yian Wang, Ziyan Xiong +3Deformable Linear Objects Robot Systems
University of Massachusetts Amherst

Sep 20, 2026, cs.ROJ/K move · Enter open · S save

RopeFormer: Cross-Trial Adaptation from Interaction History for Dynamic Rope Manipulation

Menglin Wu, Kaixiang Yao, Shangbo Luan, Masayoshi Tomizuka+1

University of California, Berkeley · Xi’an Jiaotong University · Southern University of Science and Technology+1

Dynamic Rope Manipulation Interaction History

Open paper AskPDF arXiv

Dynamic rope manipulation is highly sensitive to unknown object dynamics: the same robot motion can produce substantially different responses across ropes, while explicitly identifying the relevant physical properties is difficult. We present RopeFormer, a history-conditioned framework that uses prior task interaction as context for subsequent control. The policy retains cross-trial action-response history while keeping its weights fixed and requires no explicit online rope-parameter estimation. In matched simulation evaluations across sustained single-arm rotation, bimanual rotation, and transient whipping, retaining context improves subsequent control relative to resetting the same checkpoint, with the benefit varying across rope dynamics and observation settings. We further deploy the frozen policies on a Unitree H1-2 with previously unseen physical ropes. From T1 to T3, target-acquisition time decreases by 30.9% for Rope Swing and 33.9% for Rope Twirl, while mean Rope Whip target hits increase from 0.2 to 2.3 out of three. These results show that prior interaction can provide effective control context for dynamic deformable-object manipulation. Robot videos, code, and data are available at https://ropeformer.github.io/.

Abstract

Explore similar work

RopeFormer: Cross-Trial Adaptation from Interaction History for Dynamic Rope Manipulation

DeformX: A Versatile Co-Simulation Framework for Deformable Linear Objects

DLO-Lab: Benchmarking Deformable Linear Object Manipulations with Differentiable Physics