LIBERO-Dynamic: Benchmarking Robotic Manipulation in Dynamic Environments

Shaojie Ren; Jingwen Sun; Zezhi Liu; Hongjin Chen; Xin Jin; Wei Zou; Zhibo Chen

cs.ROOct 17, 2026

LIBERO-Dynamic: Benchmarking Robotic Manipulation in Dynamic Environments

Authors: Shaojie Ren, Jingwen Sun, Zezhi Liu, Hongjin Chen, Xin Jin, Wei Zou, Zhibo Chen

Organizations: University of Chinese Academy of Sciences · Beijing Zhongguancun Academy · Institute of Automation, Chinese Academy of Sciences · University of Science and Technology of China ·

Abstract

Most simulation benchmarks for robotic manipulation keep target objects static and treat policy inference as instantaneous. These assumptions simplify evaluation but hide how a moving target evolves during computation and how stale an observation is when an action is executed. We introduce LIBERO-Dynamic, a dynamic extension of all 40 LIBERO tasks. It drives objects through contact friction along three trajectory families and speed tiers, while advancing the physical world according to calibrated, model-specific inference latency. The same mechanism supports training by replaying verified static demonstrations and compensating robot actions online for target displacement, producing 9,000 quality-filtered dynamic trajectories. We evaluate six pretrained policies on 1,800 dynamic conditions per configuration, compare execution strategies and action horizons with fixed checkpoints, and fine-tune three architectures using dynamic and mixed data. Faster motion degrades every policy but not uniformly; execution timing and horizon can change the behavior of the same policy; and asynchronous replanning benefits from frequent feedback only when the action queue avoids underflow. Dynamic demonstrations improve each architecture under at least one recipe, with gains varying by temporal conditioning, data composition, and training budget. A small-scale physical study shows task-dependent execution-strategy differences consistent with simulation. LIBERO-Dynamic makes object motion, inference latency, and action execution explicit and controllable for evaluating and improving robot policies in moving-target manipulation.

Figures & tables

Figure 1: Overview of LIBERO -Dynamic: (a) dynamic real-world manipulation and conventional simulation evaluation; (b) latency-aware evaluation; and (c) comparisons across execution and training settings.

Benchmark	Sim. tasks	Controlled dynamics	Latency- aware	Execution comparison	Dynamic demos	Data mixture
LIBERO-Plus	40	✗	✗	✗	✗	✗
LIBERO-X	600	✗	✗	✗	✗	✗
LIBERO-Safety	75	✓	✗	✗	✓	✗
DOMINO	35	✓	✗	✗	✓	✓
DynamicVLA/DOM	–	✓	✓	✓	✓	✗
ReflexBench	6	✓	✓	✓	✓	✗

Table 1: Comparison of simulation benchmarks. Task counts exclude perturbation variants; “–” indicates that DOM v1 reports nine evaluation dimensions rather than a comparable semantic-task count. Controlled dynamics denotes prescribed within-rollout motion; data mixture compares dynamic-only and mixed static–dynamic training. Checks indicate reported features or analyses; crosses indicate not reported.

Figure 2: Benchmark dimensions, dynamic-data construction, execution protocols, and representative results. Blue denotes inference; alternating light and dark green denote actions from consecutive predictions.

Figure 3: Normalized end-effector position density in the robot base frame for original LIBERO demonstrations (left) and generated LIBERO -Dynamic demonstrations (right).

Model	(ms)	$D$	speed $=0$	Ret. (%)	0.03	0.06	0.09	All	$E_{T}$	$E_{P}$	$E_{I}$
	$\tau$		Static baseline		Dynamic SR (%)
$\pi_{0.5}$ (JAX)	60	2	96.9	59.6	64.8	57.5	51.2	57.8	.715	.727	3.14
FastWAM-Joint	64	2	98.5	41.2	49.3	39.5	33.0	40.6	.593	.606	3.70
VLA-JEPA	56	2	97.2	36.7	41.0	37.5	28.7	35.7	.566	.577	1.59
FastWAM	63	2	97.6	32.9	34.2	32.5	29.5	32.1	.503	.533	2.68
$\pi_{0}$ (JAX)	57	2	94.2	33.4	36.2	32.7	25.7	31.5	.507	.530	2.65

Table 2: Six pretrained policies under asynchronous execution at their best measured

K^{*}

on the fixed 1,800 episodes. Ret. is dynamic All SR divided by the speed

=0

result.

Figure 4:

\pi_{0.5}

(PyTorch) SR versus action horizon under (a) Sync, (b) Async, (c) RTC, and (d) VLASH. Panel (a) includes a zero-latency diagnostic; panel (b) overlays steady underflow (right axis).

Figure 5: Dynamic-data fine-tuning under asynchronous execution. (a–c) Measured

K

sweeps for three models and steady underflow (right axis). (d) Best-measured SR for each model and training recipe.

Panel of Figure 6 — Figure 6: Real-robot task sequences. Each panel shows three keyframes from one successful rollout; complete sequences are in the appendix.

Figure 7: Real-robot results at

K\in\{4,8\}

. The manipulation tasks report successful trials out of 50; whack-a-mole reports successful presses over

50\times 30

s sessions.

Appendix figures & tables15 assets

Supplementary material from the paper’s appendix.

Appendix

Check	Rejection condition
Task and rollout	non-successful BDDL goal, runtime error, assisted step, or forbidden state edit
Motion evidence	platform/fixture motion below the thresholds above
Carrier path	maximum platform path error $>0.01$ m
Target following	pre-contact object–carrier relative error $>0.08$ m
Object stability	relative tilt $>35^{\circ}$ or downward displacement $>0.04$ m
Visibility	target outside the diagnostic bird-view projection for more than 25% of sampled pre-contact steps

Appendix

Table 3: Strict physical checks used for generated demonstrations. Crossing any rejection threshold prevents the episode from entering the released dataset.

Model	$K^{*}$	linear ping-pong	Catmull-Rom spline	multi-waypoint zigzag	All
$\pi_{0.5}$ (JAX)	4	62.7	51.2	59.7	57.8
FastWAM-Joint	8	45.2	36.3	40.3	40.6
VLA-JEPA	4	39.3	33.3	34.5	35.7
FastWAM	8	34.8	29.8	31.5	32.1
$\pi_{0}$ (JAX)	8	37.5	27.5	29.5	31.5
OpenVLA-OFT	4	30.8	25.0	26.0	27.3

Appendix

Table 4: Asynchronous success rate by motion family at each model’s overall best measured eligible horizon

K^{*}

. Each family contains 600 episodes; All contains 1,800.

Figure 9: Cross-model action-horizon analysis on LIBERO -Dynamic: (a) synchronous and (b) asynchronous

K

sweeps; (c) SR by object speed at each model’s best measured asynchronous

K

Figure 10: Extended

\pi_{0.5}

(PyTorch) strategy sweeps, including two RTC settings and asynchronous underflow. Markers are measured results; smooth SR curves are visual guides.

Strategy	(ms)	$D$	0.03	0.06	0.09	All	$E_{T}$	$E_{P}$	$E_{I}$
	$\tau$		SR (%)
VLASH	104	3	65.7	54.7	46.0	55.4	.671	.689	5.61
Sync Hold (zero-lat.)	0	0	57.7	49.8	43.7	50.4	.658	.664	10.11
Async Queue	147	3	55.7	45.2	44.3	48.4	.634	.648	3.56
Sync Hold	147	3	48.7	42.5	38.0	43.1	.593	.618	9.85
RTC (fixed ${=}$ 10)	170	4	44.7	38.3	30.5	37.8	.556	.565	5.02

Appendix

Table 5: Execution-strategy results on LIBERO -Dynamic at the best measured eligible horizon for each setting (1,800 episodes per row). Sync, Async, and RTC use

\pi_{0.5}

(PyTorch); VLASH uses its official checkpoint. Zero-latency Sync is an idealized timing diagnostic.

Strategy	$K$	Steps	Path (m)	$\Delta a$	$A_{\mathrm{EEF}}$ (m/s 2 )
Async Queue	6	167.1	1.159	0.142	0.379
Sync Hold	10	183.5	1.170	0.133	0.605
RTC (fixed ${=}$ 10)	16	174.3	1.236	0.114	0.364

Appendix

Table 6: Behavioral metrics on 343 matched evaluation conditions successful under all compared strategies.

\Delta a

measures action change, and

A_{\mathrm{EEF}}

is a control-sampled acceleration estimate.

Model	FT data	Steps	Batch	Peak LR	Warm-up
$\pi_{0.5}$	Dynamic only	1,500	256	$5{\times}10^{-5}$	75
$\pi_{0.5}$	Dynamic + static	3,000	256	$5{\times}10^{-5}$	150
Joint	Dynamic only	3,000	128	$1{\times}10^{-4}$	150
Joint	Dynamic + static	3,000	128	$1{\times}10^{-4}$	150
History Joint	Dynamic only	3,000	128	$1{\times}10^{-4}$	150
History Joint	Dynamic + static	3,000	128	$1{\times}10^{-4}$	150

Appendix

Table 7: Dynamic fine-tuning configurations. Batch is the effective global batch. The mixed set is the union of the dynamic and repaired static demonstrations described above.

Model	Training data	(ms)	$D$	$K^{*}$	0.03	0.06	0.09	All	$E_{T}$	$E_{P}$	$E_{I}$
		$\tau$			SR (%)
$\pi_{0.5}$ (JAX)	Static	60	2	4	64.8	57.5	51.2	57.8	.715	.727	3.14
	Dynamic only			4	70.5	63.3	45.0	59.6	.726	.719	3.38
	Dynamic + static			4	81.5	70.8	55.7	69.3	.788	.779	4.50
FastWAM-Joint	Static	64	2	8	49.3	39.5	33.0	40.6	.593	.606	3.70
	Dynamic only			4	64.7	48.5	35.0	49.4	.651	.671	2.51

Appendix

Table 8: Results for three training recipes under asynchronous execution. Each row uses its best measured horizon

K^{*}

and 1,800 episodes.

Model	Training	Sync (%)	Async (%)
Single-frame Joint	Static base	43.0	40.6
Single-frame Joint	Dynamic-only FT	52.0	45.5
History Joint	Static base	43.5	43.8
History Joint	Dynamic-only FT	62.4	67.1
$\pi_{0.5}$	Official (static)	59.3	55.9
$\pi_{0.5}$	Mixed FT	65.4	63.8

Appendix

Table 9: Fixed-

K

comparison before and after dynamic-data fine-tuning on LIBERO -Dynamic (

K{=}8

; 1,800 episodes per configuration). Sync is zero-latency, and Async uses calibrated latency.

Model	Exec.	$K$	Conv. /50	Smooth /50	Zigzag /50	Whack presses
$\pi_{0.5}$	Async	4	30	16	18	391
	Async	8	23	12	16	374
	RTC	4	28	13	15	408
	RTC	8	20	10	14	393
FastWAM-Joint	Async	4	19	8	12	369
	Async	8	16	7	10	348

Appendix

Table 10: Real-robot results for six model–execution configurations at

K{\in}\{4,8\}

. Conveyor and the two car-pursuit conditions report successful trials out of 50; whack-a-mole reports total successful presses across

50{\times}30

s sessions.

Model	Execution	Input	$H$	Denoising	Evaluated $K$
$\pi_{0.5}$	Async	Current	16	10	4, 8
$\pi_{0.5}$	RTC	Current + RTC	16	10	4, 8
Joint	Async	Current	32	1	4, 8
Joint	RTC	Current + RTC	32	10	4, 8
History Joint	Async	5 frames + 16 actions	16 †	1	4, 8
History Joint	RTC	History + RTC	16 †	10	4, 8

Appendix

Table 11: Deployment configuration for the real-robot results in Table 10 . All six settings run at 15 Hz and are evaluated at

K{\in}\{4,8\}

. “History” denotes the five-frame input and 16-action clean prefix.

Figure 11: Conveyor handover: seven keyframes from one successful episode.

Figure 12: Car pursuit with a smooth-polynomial trajectory: seven keyframes from one successful episode.

Figure 13: Car pursuit with a zigzag trajectory: seven keyframes from one successful episode.

Figure 14: Whack-a-mole: seven keyframes from one successful episode.

Dynamic Robotic Manipulation

Dynamic robotic manipulation studies how robots act on moving or deformable objects and respond to fast-changing physical interactions.

See all 89 papers

DynamicManip: Enabling Dynamic Manipulation from a Single Static Demonstration
Aug 2, 2026Haoran Liao, Pengyue Wang, Shuoyu Chen +10Robot Policy Learning Synthetic Data Generation for Robotics
1Sun Yat-sen University · 4Jiangxing Intelligence · 2Stanford University +1
LIBERO-Industry: Benchmarking Robotic Manipulation under Industrial Process Constraints
Oct 12, 2026Ming Shin Cindy Sung, Duan Leijiahao, Haoxiang Wang +9Robotic Manipulation Robot Manipulation Benchmarks
Institute of AI for Engineering, Tongji University · The Chinese University of Hong Kong · University of Aberdeen +1
LIBERO-Agent: Evaluating General-Purpose Agents for Direct Embodied Manipulation
Sep 30, 2026Zijie Diao, Yitong Chen, Sicheng Xie +7Long-Horizon Robotic Manipulation Contact-Rich Robotic Manipulation
Institute of Trustworthy Embodied AI, Fudan University · Shanghai Innovation Institute

Aug 2, 2026J/K move · Enter open · S save

DynamicManip: Enabling Dynamic Manipulation from a Single Static Demonstration

Haoran Liao, Pengyue Wang, Shuoyu Chen, Kehan Cheng+9

1Sun Yat-sen University · 4Jiangxing Intelligence · 2Stanford University+1

Robot Policy Learning Synthetic Data Generation for Robotics Imitation Learning Dynamic Robotic Manipulation Robot Manipulation Benchmarks

Open paper AskPDF arXiv

Dynamic manipulation is a critical capability for robots operating in complex and dynamic environments, where robots must interact with objects that are moving or require rapid adjustments. However, learning models for dynamic manipulation tasks face two major challenges: (1) the combinatorial complexity of dynamic scenarios leads to substantial data requirements, and (2) rapid variations in dynamics require real-time and accurate policy execution. In this paper, we propose DynamicManip to address these challenges through an efficient data augmentation pipeline and a low-latency imitation policy. We first propose a static-to-dynamic augmentation pipeline that synthesizes diverse dynamic manipulation demonstrations from a single static demonstration. Second, we introduce a dynamic-aware adaptive policy that adjusts its inference frequency according to task dynamics, enabling responsive and effective dynamic manipulation. Third, we build a dynamic manipulation benchmark, which includes diverse dynamic tasks with an automatic evaluation system for scalable and consistent assessment. Extensive experiments in both simulation and the real world demonstrate that DynamicManip not only provides significant improvements in data efficiency but also achieves better performance in dynamic manipulation tasks, with a mean success rate 18.4 percentage points higher and policy-query latency 32.9% lower.

Abstract

Dynamic Robotic Manipulation

Explore similar work

DynamicManip: Enabling Dynamic Manipulation from a Single Static Demonstration

LIBERO-Industry: Benchmarking Robotic Manipulation under Industrial Process Constraints

LIBERO-Agent: Evaluating General-Purpose Agents for Direct Embodied Manipulation