Reliable 3D dynamic perception requires models that can anticipate motion beyond predefined categories, yet progress is hindered by the scarcity of dense, high-quality motion annotations. While self-supervision on unlabeled real data offers a path forward, empirical evidence suggests that scaling unlabeled data fails to close the performance gap due to noisy proxy signals. In this paper, we propose learning robust real-world motion priors entirely from scalable simulation. We introduce SynFlow, a data generation pipeline for large-scale synthetic LiDAR scene flow. Unlike prior works that prioritize sensor-specific realism, SynFlow employs a motion-oriented strategy to synthesize diverse kinematic patterns across 4,000 sequences (∼940k frames), termed SynFlow-4k. This represents a 34× scale-up in annotated volume over existing real-world benchmarks. Our experiments demonstrate that SynFlow-4k provides a highly domain-invariant motion prior. In a zero-shot regime, models trained only on our synthetic data generalize across multiple real-world benchmarks, comparable to in-domain supervised baselines on nuScenes and outperforming state-of-the-art methods on TruckScenes by 31.8%. Furthermore, SynFlow-4k serves as a label-efficient foundation: fine-tuning with only 5% of real-world labels surpasses models trained from scratch on the full available budget. We open-source the pipeline and dataset to facilitate research in generalizable 3D motion estimation. More detail can be found at https://kin-zhang.github.io/SynFlow.
Figures & tables
Figure 1: Scaling up LiDAR Scene Flow with Synthetic Data. We present SynFlow, a data generation pipeline leveraging the CARLA simulator to synthesize diverse, perfectly labeled LiDAR scene flow data (center). While real-world datasets are often constrained by high annotation costs and limited scenario diversity, SynFlow provides a scalable source of dense, noise-free supervision for learning robust motion priors. As shown in the results (right), models trained on SynFlow dataset achieve strong zero-shot generalization on real-world benchmarks and significantly outperform in-domain baselines when fine-tuned on a small subset of real data.
Figure 2: Overview of our SynFlow pipeline and dataset examples. Left: A CARLA world provides diverse road topologies; we construct a route bank using topology-aware coverage to ensure broad spatial exploration and execute rollouts under Traffic Management (TM). Middle: our procedural data engine instantiates an ego vehicle with configurable LiDAR, spawns surrounding agents with controllable policies, and runs synchronized simulation steps; Right: Representative frames showing raw LiDAR point clouds alongside corresponding dense scene flow labels, spanning diverse traffic interactions and motion regimes.
Split
#Annotated Frames
Composition
Scenarios
1k
271,148
Town06, 07, 10
arterials, complex junctions, rural roads
2k
449,407
Town01–05
roundabouts, multi-lane intersections
3k
720,555
1k + 2k
(1k + 2k)
4k
939,083
2k + Town12
mixed zones (urban/suburban/rural)
Table 1: Summary of SynFlow-4k data splits. Scaling annotated volume via rollouts across CARLA towns and route banks. Training sets comprise mixed 32/64-beam LiDAR sequences.
Methods
Supervision
Dynamic Bucket-Normalized ↓
Three-way EPE (cm) ↓
Mean
CAR
OTHER
PED.
VRU
Mean
FD
FS
BS
Ego Motion Flow
–
1.000
1.000
1.000
1.000
1.000
12.34
35.94
1.07
0.00
Self-supervised (100% unlabeled real-world data)
SeFlow [ 45 ]
100% unlab.
0.544
0.396
0.635
0.726
0.419
8.19
16.15
3.97
4.45
VoteFlow [ 18 ]
100% unlab.
0.538
0.355
0.605
0.780
0.410
7.80
15.65
3.51
4.24
SeFlow++ [ 43 ]
100% unlab.
0.509
0.327
0.583
0.716
0.409
6.13
14.59
1.96
1.86
Table 2: Performance on the nuScenes validation set. Supervision : unlab. = self-supervised on unlabeled real sequences; lab. = supervised on available (20%) labeled real data; synth. = supervised on SynFlow synthetic data only (no real data). Bold denotes the best result in each column.
Methods
Supervision
Dynamic Bucket-Normalized ↓
Three-way EPE (cm) ↓
Mean
CAR
OTHER
PED.
VRU
Mean
FD
FS
BS
Ego Motion Flow
–
1.000
1.000
1.000
1.000
1.000
61.43
184.30
0.00
0.00
Self-supervised (100% unlabeled real-world data)
SeFlow [ 45 ]
100% unlab.
0.681
0.494
0.752
0.886
0.591
37.41
103.97
1.56
6.68
VoteFlow [ 18 ]
100% unlab.
0.680
0.517
0.737
0.895
0.573
36.47
103.48
1.29
4.64
SeFlow++ [ 43 ]
100% unlab.
0.653
0.519
0.738
0.860
0.497
33.35
95.62
1.20
3.23
Table 3: Performance on the TruckScenes validation set. TruckScenes features large commercial vehicles with distinct LiDAR configurations unseen during SynFlow pre-training, providing a stringent test of cross-domain zero-shot generalization. Notation follows Tab. 2 .
Figure 3: Zero-shot scaling performance (1k–4k sequences). Evaluation on Aeva (a) and TruckScenes (b) using Dynamic Bucket-Normalized EPE (lower is better). Solid blue line indicates the overall mean; dashed lines represent per-category breakdowns.
Methods
Real labels
Mean
CAR
OTHER
PED.
VRU
nuScenes validation set
Δ Flow [ 46 ]
20%
0.216
0.138
0.219
0.327
0.181
SynFlow-4k (Zero-shot)
0%
0.242
0.177
0.307
0.300
0.183
SynFlow-4k + FT
5%
0.201
0.151
0.240
0.272
0.139
SynFlow-4k + FT
10%
0.175
0.127
0.201
0.253
0.119
SynFlow-4k + FT
20%
0.157
0.110
0.173
0.247
0.098
Table 4: Fine-tuning scaling on nuScenes (upper) and TruckScenes (lower). Comparison between from-scratch supervised baselines and SynFlow-4k across varying real-world label budgets. Zero-shot (0%) denotes purely synthetic supervision; +FT rows indicate fine-tuning on the specified real-world fraction. Bold denotes best per column.
Training data
Dynamic Bucket-Normalized ↓
Three-way EPE (cm) ↓
Mean
CAR
OTHER
PED.
VRU
Mean
FD
FS
BS
UniFlow only
0.303
0.112
0.359
0.398
0.344
6.65
13.17
5.16
1.62
SynFlow-4k only
0.288
0.166
0.430
0.248
0.309
9.08
21.41
4.88
0.94
Both
0.263
0.102
0.401
0.251
0.298
5.95
13.15
3.85
0.84
Table 5: Complementarity of SynFlow-4k and UniFlow. Zero-shot evaluation on the Aeva dataset. UniFlow [ 16 ] is trained on merged real-world data (Argo-v2 [ 36 ] , nuScenes, Waymo). Combining synthetic and real-world sources consistently outperforms individual training.
P1: Topology
P2: Speed
Dynamic Bucket-Normalized ↓
Three-way EPE (cm) ↓
Mean
CAR
OTHER
PED.
VRU
Mean
FD
FS
BS
0.346
0.225
0.512
0.311
0.334
18.96
52.34
3.51
1.02
✓
0.331
0.211
0.487
0.282
0.342
18.45
51.25
3.10
1.01
✓
0.330
0.215
0.501
0.281
0.324
17.89
50.21
2.82
0.65
✓
✓
0.319
0.201
0.488
0.270
0.317
17.11
48.17
2.52
0.66
Table 6: Ablation of data generation design choices (1k sequences trained), zero-shot evaluated on Aeva dataset. “Topology Coverage (Policy 1)” toggles greedy lane-segment coverage route bank; otherwise, random-start fixed-length routes are used. “Speed Regime (Policy 2)” toggles whether highway-structured towns (Town04, Town06, Town07) are included; otherwise only urban towns are used.
Backbone
Dynamic Bucket-Normalized ↓
Three-way EPE (cm) ↓
MEAN
CAR
OTHER
PED.
VRU
Mean
FD
FS
BS
Ego Motion Flow
1.000
1.000
1.000
1.000
1.000
44.31
132.00
0.92
0.00
Trained on SynFlow-4k
DeFlow
0.426
0.249
0.514
0.448
0.500
10.99
25.85
6.60
0.84
Flow4D
0.308
0.173
0.452
0.286
0.320
9.37
21.87
6.00
0.94
Δ Flow
0.288
0.166
0.430
0.248
0.309
9.08
21.41
4.88
0.94
Table 7: Ablation of backbone architectures. Zero-shot evaluation on Aeva after training on SynFlow-4k dataset. The consistent improvement across different estimators demonstrates the backbone-agnostic transferability of our synthetic supervision.
Figure 4: Dataset visualization. Top: SynFlow-4k samples under 64-beam (row 1) and 32-beam (row 2) configurations, spanning city, roundabout, highway, and merging scenarios. Per-point scene flow labels are rendered as colored vectors. Direction is encoded as hue, and magnitude as saturation. Bottom: representative samples from real-world datasets (nuScenes and TruckScene).