Vision-language-action (VLA) policies have become a promising paradigm for end-to-end robot control, but their effectiveness still depends heavily on large volumes of demonstration data. In manipulation tasks, this dependence is compounded by structured spatiotemporal uncertainty in the action space: multiple nearby actions or trajectories may successfully solve the same task, while semantically equivalent demonstrations often unfold at different speeds. Standard pointwise supervision treats each demonstrated action as an exact target, imposing an unnecessary learning burden under such uncertainty. We present SURE, a data-efficient VLA training framework that explicitly models both spatial and temporal uncertainty. SURE augments a VLA backbone with a lightweight transposed-convolutional probabilistic action decoder that captures spatial uncertainty through local correlations among neighboring action bins. It further introduces Soft-DTW alignment supervision to accommodate local timing variations between predicted and demonstrated action chunks. Together, these two components reduce sensitivity to demonstrator-specific poses and execution timing, enabling more data-efficient action learning. We evaluate SURE on LIBERO, RoboTwin 2.0, and real-world manipulation tasks. Even with only 10% of the demonstration data or 20% of the training steps, SURE achieves manipulation success rates that match or surpass baseline VLAs.
Figures & tables
Figure 1: Overview of SURE. (a) Given observations and a language instruction, an OpenVLA-7B backbone produces action tokens that are decoded into action chunks. (b) Our probabilistic action decoder models translation and rotation as discretized distributions and recovers continuous controls with expectation-based decoding, capturing spatial uncertainty and multimodality. (c) During training, Soft-DTW alignment supervision aligns predicted and demonstrated action chunks that progress at different speeds, addressing temporal uncertainty in robot demonstrations.
Figure 2: Architecture of the probabilistic action decoder. Action tokens Ht∈Rk×da×dVLM are first fused by a grouped 1×1 convolution and then split into translation, rotation, and gripper branches. Translation and rotation use two Deconv Blocks to upsample an ordered bin axis from 1→R/4→R . The resulting per-axis logits over R bins are normalized by softmax and decoded into continuous actions by expectation. Each Deconv Block combines depthwise transposed convolution with pointwise residual layers to encode local correlations among neighboring bins. The gripper action is decoded by a lightweight MLP.
Methods
Steps(BS)
Spatial
Object
Goal
Long
Avg.
Diffusion Policy † ( Chi et al., 2023 )
100-250k(64)
78.3
92.5
68.3
50.5
72.4
Octo † ( Ghosh et al., 2024 )
100-250k(64)
78.9
85.7
84.6
51.1
75.1
OpenVLA † ( Kim et al., 2025b )
50-150k(64)
84.7
88.4
79.2
53.7
76.5
GR00T-N1 † ( NVIDIA et al., 2025 )
/ (128)
94.4
97.6
93.0
90.6
93.9
π0 † ( Black et al., 2025 )
/ (128)
96.8
98.8
95.8
85.2
94.2
OpenVLA-OFT ( Kim et al., 2025a )
10k(48)
94.4
92.6
92.6
68.8
87.1
Table 1: Success rates (%) on the four LIBERO task suites. Training steps and batch sizes (BS) are reported when available. SURE is trained for 10k optimization steps and achieves the highest average success rate. Bold values indicate the best result in each column. † denotes results averaged over three random seeds, all other results are obtained using a single evaluation seed.
Methods
Training steps(BS)
Beat Block Hammer
Click Alarm Clock
Lift Pot
Place Fan
Stack Bowls Two
Avg.
Diffusion Policy
20-75k(128)
42.0
61.0
39.0
3.0
61.0
41.2
OpenVLA-OFT
100k(8)
18.0
85.0
93.0
25.0
75.0
59.2
RDT ( Liu et al., 2025 )
10k(64)
77.0
61.0
72.0
12.0
76.0
59.6
π0
30k(32)
43.0
63.0
84.0
20.0
91.0
60.2
SURE
20k(16)
55.0
100.0
98.0
26.0
82.0
72.2
Table 2: Success rates (%) on five RoboTwin 2.0 tasks under the Easy evaluation setting. All methods are trained on the clean subset with a separate checkpoint for each task. Training steps and batch sizes (BS) are reported for each method. Bold values indicate the best result in each column.
Methods
Water Bottle
AAA Battery
Screwdriver
Toy Duck
Stack Bowls Two
Pour Water
Avg.
OpenVLA-OFT
40
35
55
45
20
55
41.7
SURE
55
60
70
95
25
40
57.5
Table 3: Success rates (%) on six real-world manipulation tasks. Each method is evaluated over 20 trials per task. OpenVLA-OFT and SURE are trained for 50k and 20k optimization steps, respectively. Bold values indicate the best result in each column.
Stochastic
Methods
Params
FLOPs
Latency
No
OFT
151.12M
2.42G
0.37ms
Yes
π0.5 (JAX)
430.10M
79.13G
19.88ms
SURE
16.93M
10.96G
1.57ms
SURE ∗ (JAX)
16.93M
10.96G
0.69ms
Table 4: Comparison of action head complexity and inference efficiency. Stochastic indicates whether the action head explicitly models the action distribution. Params denotes the number of parameters in the action head. Latency denotes the mean time required by the action head to decode one action chunk, excluding the VLA backbone. SURE ∗ denotes our JAX variant used for a framework-matched comparison with π0.5 .
Figure 3: Success rates (%) on LIBERO-Spatial using different fractions of the available training demonstrations. Both methods are trained for 10k optimization steps.
Variant
Steps
Prob
DTW
Success Rate
SURE
10k
✓
✓
99.0
w/o DTW
10k
✓
–
97.4
w/o Prob
10k
–
✓
96.8
w/o Both
10k
–
–
94.4
Table 5: Subtractive ablation on LIBERO-Spatial. Prob denotes the probabilistic action decoder, and DTW denotes Soft-DTW alignment supervision. All variants are trained for 10k steps.
Appendix figures & tables4 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 4: Representative SURE action-distribution visualizations on LIBERO-Spatial, LIBERO-Object, LIBERO-Goal, and LIBERO-Long, from top to bottom.
Figure 5: Representative SURE joint-action distribution visualizations on RoboTwin 2.0: Beat Block Hammer, Click Alarm Clock, Place Fan, and Stack Bowls Two, from top to bottom.
Figure 6: Real-world experimental setup with LeRobot SO-101 robotic arms. A front camera observes the workspace, while a wrist-mounted camera captures fine-grained manipulation details.
Figure 7: Real-world manipulation rollouts. Columns show the temporal progression from t0 to t4 , while rows correspond to the six tasks. To provide clearer visualizations than the low-resolution front camera used by the policy, these frames are captured by an auxiliary camera placed alongside the front camera. Consequently, the robot arms are not centered in the images.