FlowS: One-Step Motion Prediction via Local Transport Conditioning

arXiv:2604.26065 · cs.RO · Submitted 2026-04-28 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.

Rosa: We're looking at the paper: "FlowS: One-Step Motion Prediction via Local Transport Conditioning".

Dev: Here's the gist of the paper. Generative motion prediction must satisfy three simultaneous requirements for real-world autonomy: high accuracy, diverse multimodal futures, and strictly bounded latency The single most informative sentence about the paper's subject and finding is that local transport conditioning makes one-step generative motion prediction practical for safety-critical autonomy.

Rosa: The core problem addressed Motion prediction for autonomous driving must satisfy three simultaneous requirements: high accuracy, multimodal coverage, and strictly bounded latency The essential challenge is clear: given the recent scene with spatial positions and movement dynamics, predict where all agents will be within the next few seconds.

Dev: Diffusion models meet the first two but violate the third, requiring tens to hundreds of denoising steps.

Rosa: This observation reveals a strategy that we believe is fundamental to few-step generative prediction: Single-step integration is accurate when the transport problem is conditioned to be local.

Title and authors: Rosa: Now we look at the title again: FlowS: One-Step Motion Prediction via Local Transport Conditioning. It really tells you exactly what the paper is trying to solve.

Dev: It’s about taking motion prediction, which usually requires a lot of steps, and making it one step by conditioning it on local transport problems so we can achieve high accuracy without the latency penalty.

Taro: I wonder if they are really conditioning it on *local* transport in a general sense, or just local to the agent’s immediate vicinity. That’s where I want to probe.

Rosa: They explain that when you condition the base distribution to lie near plausible futures, you reduce the problem from discovering a global mode and traversing a long displacement to just short-range refinement.

Dev: That sounds like they are using the prior as a way to prune the search space dramatically before the main flow model even has to work hard.

Taro: Pruning is good, but I need to know if that pruning isn't just hiding errors in low-probability scenarios. What happens when the world misbehaves outside those anchors?

Rosa: They address that by using a scene-conditioned learned prior that emits K calibrated anchor trajectories per agent, each already near a plausible future, which converts mode discovery into local correction.

Dev: So instead of sampling from a huge Gaussian noise distribution and hoping for the right path, the AI starts with K options that are already somewhat correct? That should be much more efficient.

Taro: But if those K anchors don't cover the right maneuver, we’re back to square one, and I don't want to waste time on something that fails spectacularly in a real-world scenario.

Rosa: The paper shows that this prior reduces mean transport distance by thirty-six percent compared to starting from a standard Gaussian base sample Z ∼ N (zero I), which is about one point two three versus zero point seven eight, according to Figure five.

Dev: That thirty-six percent reduction in transport distance is significant because it directly tackles the long-range displacement issue that causes those massive discretization errors in a single step prediction.

Taro: So the prior helps by shortening the path, but what about the actual velocity estimation? The paper mentions that without this prior, the displacement field alone needs a mean velocity of one point seven two to bridge that full Gaussian-to-data gap.

Rosa: That's because without anchors, you have to learn everything from scratch—both where the agent is going and how fast it’s moving. The prior helps by eliminating the mode discovery burden while the displacement field eliminates discretization error on that short transport.

Dev: It’s a dual mechanism working together; one component handles distance reduction, and the other handles fidelity of movement estimation during that short path. That sounds like a very solid engineering approach for real-time performance.

Taro: I’m still focused on the training signal itself; how do they ensure that these two components actually cooperate during the learning phase without introducing instability?

Rosa: They use Conditional Flow Matching, or CFM, which learns a time-indexed flow model vphi by matching it to a known velocity target along a family of conditional probability paths. This is trained on an optimal transport straightline path between Z and Y.

Dev: So the training target is fixed as the empirical displacement from the base sample to the ground truth future, constant along each Optimal Transport path, which gives them something stable to match against.

Taro: That sounds like a good anchor for stability, unlike some prior self-consistency methods that can suffer from high-variance bootstrap signals when they are on curved diffusion paths.

Rosa: Exactly. The paper notes that anchoring the displacement field at learned priors along straightline paths yields a stable, low-variance training target, which helps stabilize the prior formulation compared to other ways of doing it.

Dev: So the key finding here is that this specific combination of conditioning and consistency creates a very stable signal for training the flow model. It’s less sensitive to noise during the learning process.

Taro: That stability is crucial for deployment; we need models that are robust when they encounter unexpected inputs, not just when they are in a perfectly straight-line training environment.

The paper's summary: Rosa: So, summarizing the core of FlowS: One-Step Motion Prediction via Local Transport Conditioning is that diffusion models struggle with latency because they need many steps to be accurate and diverse, but they found a way around this by conditioning the problem.

Dev: They identified that single-step integration is accurate if the underlying transport problem is local, meaning we only need to worry about short-range corrections instead of long-range discoveries.

Taro: The mechanism they use for this localization is twofold: first, a scene-conditioned learned prior that gives us K anchor trajectories already close to plausible futures.

Rosa: That prior converts the mode discovery problem into a local correction task, essentially telling the model where it’s probably going to be before it even starts predicting.

Dev: And then they add the second part: a step-consistent displacement field that ensures that even though we only take one update at inference time, that single step still respects the accuracy of multi-step dynamics during training.

Taro: So you have a starting point based on plausible futures, and then a mechanism to ensure the next move is physically consistent with what would happen over several steps.

Rosa: Precisely. The synergy between these components—the prior reducing transport distance and the field absorbing curvature—is what leads to state of the art performance on the Waymo Open Motion Dataset.

Dev: They achieved a soft mAP of zero point four eight zero four and an mAP of zero point four seven zero three, both with single-step inference running at about seventy-five frames per second, which is quite fast for real-time use cases.

Taro: That speed combined with those scores shows that this technique can actually be practical for safety critical autonomy when we need fast answers.

Rosa: It confirms the idea that conditioning on local transport makes one-step generative motion prediction practical for safety critical autonomy, which is the central finding of FlowS: One-Step Motion Prediction via Local Transport Conditioning.

The paper's improvements: Dev: Let's talk about what they actually improved in terms of methodology. They introduced a few specific changes to how they handle the flow matching process.

Rosa: The first improvement is using the learned prior instead of starting from a standard Gaussian base sample Z ∼ N (zero I) <ref:2604.26065#pg1>. As we saw in Figure three, this drastically reduces transport distance from one point two three down to zero point seven eight, which is a thirty-six percent reduction in mean transport distance.

Taro: Thirty-six percent is substantial; that means the AI doesn't have to travel as far across the state space before it starts making meaningful corrections, which directly lowers the error accumulation.

Dev: And then they implemented a step-consistent displacement field, spsi(Xt, t, d, c) which enforces semigroup self-consistency by ensuring that one two-dimensional update matches two composed d updates.

Taro: That’s the part that addresses discretization error in the flow dynamics; it makes sure the single step is actually a good approximation of what would happen over several steps.

Rosa: And they showed how these two components work together: the prior reduces transport distance by thirty-six percent, while the displacement field absorbs marginal curvature in vphi via semigroup consistency.

Dev: That dual mechanism is key; one component handles the distance reduction, and the other handles the cumulative effect of curvature in vphi during training. It stabilizes the bootstrap signal during training and ensures that when you take that single step at inference time, it realizes the full transport in one pass.

Taro: So they are not just improving one part; they’re fixing two independent sources of error: reducing the distance traveled and absorbing how quickly things change over time in a curved path.

Rosa: Exactly. They also showed that adding just the prior alone improves mAP by plus zero point zero zero eight two, and adding the step-consistent field alone yields an improvement of plus zero point zero zero five six on the validation set for component ablation studies, which is quite telling about their individual contributions.

Dev: It’s a solid finding because it quantifies exactly how much each piece contributes to the final result so you can see where the gains are coming from.

Taro: I'm still looking at those limitations they mentioned—the paper says the learned prior’s K equals six mixture may not capture all tail distribution maneuvers, especially for cyclist U-turns. That tells us we still have a gap in coverage that needs work before this is ready for mission critical use.

Rosa: That limitation is explicitly stated in their evaluation on the Waymo Open Motion Dataset; they say transfer to datasets with different map representations or shorter horizons would require encoder adaptation and prior recalibration.

Conclusion: Dev: So, wrapping up the FlowS: One-Step Motion Prediction via Local Transport Conditioning paper, the main implication is that we can achieve high accuracy and low latency for motion prediction by using local transport conditioning.

Rosa: They did this by creating a system where the learned prior anchors plausible futures, drastically reducing transport distance and then using a consistent displacement field to ensure single-step accuracy is maintained.

Taro: For me, it means that we have a way to build systems that can make fast decisions on the road without needing massive amounts of iterative computation for every decision.

Dev: It moves the problem from requiring hundreds of denoising steps down to one deterministic forward pass at about seventy-five frames per second, which is what we need for real-time systems.

Rosa: We're seeing strong performance on the Waymo Open Motion Dataset with mAP of zero point four seven zero three and soft mAP of zero point four eight zero four, showing that this method works effectively in practice when tested outside the lab environment as well.

Taro: I just hope they can expand beyond those initial evaluation sets to show that this robustness holds up when the world gets genuinely messy and unpredictable.

Dev: That's the next challenge; moving from a single-step prediction to something that handles more complex interaction scenarios with joint prediction, which is a natural next step for this research.

Rosa: So, FlowS: One-Step Motion Prediction via Local Transport Conditioning provides a concrete method for achieving one-step generative motion prediction practical for safety critical autonomy by grounding the displacement field at learned priors along straightline paths.

Leandro Di Bella, Bruno Cornelis, Adrian Munteanu

Vrije Universiteit Brussel · imec

cs.RO

Submitted: 2026-04-28

Updated: 2026-10-28

Comments: 8 pages

Journal ref: IEEE Robotics and Automation Letters, vol. 11, no. 12, pp. 13805-13812, Dec. 2026

DOI: 10.1109/LRA.2026.3739561

License: http://creativecommons.org/licenses/by-nc-nd/4.0/

Importance score: 89/100

The gist: Generative motion prediction must satisfy three simultaneous requirements for real-world autonomy: high accuracy, diverse multimodal futures…

Key concepts

Local Transport Conditioning
This strategy makes motion prediction practical by ensuring the underlying transport problem is local. Instead of predicting a long, complex movement in one go, the model is guided to only perform short corrections near plausible future trajectories. This reduces discretization errors and allows a single step to be highly accurate.
Scene-Conditioned Learned Prior
This component replaces the standard random starting point with learned anchor trajectories already close to realistic futures. It converts the difficult task of discovering the correct overall movement mode into a simpler task of local correction, significantly reducing the transport distance required for prediction.
Step-Consistent Displacement Field (spsi)
This field enforces a mathematical consistency rule: two small sequential steps must approximate one larger step. This mechanism ensures that even though only one step is taken during inference, it inherits the accuracy of multi-step predictions, stabilizing training and reducing error caused by marginal curvature.

Terminology

Summary

Generative motion prediction must satisfy three simultaneous requirements for real-world autonomy: high accuracy, diverse multimodal futures, and strictly bounded latency The single most informative sentence about the paper's subject and finding is that local transport conditioning makes one-step generative motion prediction practical for safety-critical autonomy.

The core problem addressed

Motion prediction for autonomous driving must satisfy three simultaneous requirements: high accuracy, multimodal coverage, and strictly bounded latency The essential challenge is clear: given the recent scene with spatial positions and movement dynamics, predict where all agents will be within the next few seconds. Diffusion models meet the first two but violate the third, requiring tens to hundreds of denoising steps. p. 0

Where the paper says it

FlowS: One-Step Motion Prediction via Local Transport Conditioning Leandro Di Bella1 , Bruno Cornelis1 , and Adrian Munteanu1 Member, IEEE Abstract—Generative motion prediction must satisfy three simultaneous requirements for real-world autonomy: high accuracy, diverse multimodal futures, and strictly bounded latency. Diffusion models meet the first two but violate the third, requiring tens to hundreds of denoising steps.

This observation reveals a strategy that we believe is fundamental to few-step generative prediction: Single-step integration is accurate when the transport problem is conditioned to be local. p. 0
Where the paper says it

Efforts to speed up sampling, such as DiffusionDrive [8] (truncated schedules from learned anchors) and BridgeDrive [9] (diffusion bridges with matched PF-ODE sampling), have reduced steps but still require multiple refinement cycles and cannot guarantee single-cycle latency. This observation reveals a strategy that we believe is fundamental to few-step generative prediction: Single-step integration is accurate when the transport problem is conditioned to be local.

The conditioning strategy

Conditioning the base distribution to lie near plausible futures reduces the transport to a short local correction, the regime in which single-step integration is accurate. FlowS implements this with two complementary mechanisms:

  1. Scene-conditioned learned prior. p. 0

    Where the paper says it

    We instantiate this local transport conditioning in FlowS, a conditional flow matching framework with two mechanisms. First, an online, scene-conditioned learned prior emits K calibrated anchor trajectories per agent, each already near a plausible future, converting mode discovery into local correction.

    Rather than starting from N (0, I), we learn an online, scene-conditioned multimodal prior that emits K calibrated anchor trajectories per agent, each already near a plausible future This converts the global mode-discovery problem into local correction. p. 0
    Where the paper says it

    Rather than starting from N (0, I), we learn an online, scene-conditioned multimodal prior that emits K calibrated anchor trajectories per agent, each already near a plausible future. This converts the global mode-discovery problem into local correction, satisfying the locality condition.

  2. Step-consistent displacement field. p. 0

    Where the paper says it

    This converts the global mode-discovery problem into local correction, satisfying the locality condition. • Step-consistent displacement field.

    p. 0
    Where the paper says it

    This converts the global mode-discovery problem into local correction, satisfying the locality condition. • Step-consistent displacement field.

    We train a semigroup-consistent transport field that enforces T∆t ◦ T∆t ≈ T2∆t, so that a single step faithfully approximates multi-step dynamics. p. 0
    Where the paper says it

    • Step-consistent displacement field. We train a semigroup-consistent transport field that enforces T∆t ◦ T∆t ≈ T2∆t, so that a single step faithfully approximates multi-step dynamics.

The training signal is built

The training involves Conditional Flow Matching (CFM) which learns a time-indexed flow model vϕ by matching it to a known velocity target along a family of conditional probability paths. A convenient choice is the optimal transport (OT) straightline path between Z and Y. p. 2

Where the paper says it

CFM learns a time-indexed flow model vϕ by matching it to a known velocity target along a family of conditional probability paths {pt(· |Z, Y )}t∈[0,1] that smoothly interpolate between p0 (at t=0) and pdata (at t=1): LCFM = EZ∼p0, Y ∼pdata t∼U[0,1] vϕ Xt, t, c − ut Xt |Z, Y 2 2 . (1) A convenient choice is the optimal transport (OT) straightline [3], [10] path between Z and Y : Xt = (1 − t)Z + t Y, ut Xt |Z, Y = Y − Z, (2) which yields a time-independent target velocity ut and closed-form supervision.

In a one-step regime, the flow is applied once from the base sample (t=0): Ypred = Z + vϕ(Z, 0, c). p. 2
Where the paper says it

Substituting (2) into (1) gives: LCFM = EZ,Y,t vϕ (1−t)Z +tY, t, c −(Y −Z) 2 2 . (3) In a one-step regime, the flow is applied once from the base sample (t=0): Ypred = Z + vϕ(Z, 0, c), (4) C.

The prior network defines the K-component mixture distribution over anchor trajectories; at inference, rather than sampling a single component, we evaluate all K modes in parallel and use the mixture weights πk as initial confidence scores for downstream selection via NMS. p. 3
Where the paper says it

A mixture calibration loss aligns predicted weights π with soft targets πˆk = softmax(−NLLk/τmix) via KL divergence: Lmix = DKLπˆ ∥ π = XK k=1 πˆk log(ˆπk/πk). (11) The prior defines a K-component mixture distribution over anchor trajectories; at inference, rather than sampling a single component, we evaluate all K modes in parallel and use the mixture weights πk as initial confidence scores for downstream selection via NMS.

Where the gains come from

The synergy between components is key: The prior reduces mean transport distance by 36% (0.78 vs. 1.23), while the displacement field alone must bridge the full Gaussian-to-data gap (mean velocity 1. p. 6

Where the paper says it

23), but a single Euler step still accumulates discretization error without scale-aware correction. Conversely, the displacement field alone must bridge the full Gaussian-to-data gap (mean velocity 1.

72). Together, the prior eliminates the mode-discovery burden while the displacement field eliminates discretization error on the resulting short transport. p. 6
Where the paper says it

72), making its self-consistency target noisy. Together, the prior eliminates the mode-discovery burden while the displacement field eliminates discretization error on the resulting short transport (mean velocity 1.

The semigroup constraint distills multi-step accuracy into a single step, resulting in a ∆mAP = 0. p. 5
Where the paper says it

latency (red, right axis) across 1–16 inference steps. The semigroup constraint distills multi-step accuracy into a single step: ∆mAP = 0.

0017 with 9. p. 5
Where the paper says it

The semigroup constraint distills multi-step accuracy into a single step: ∆mAP = 0. 0017 with 9.

5× speedup. p. 5
Where the paper says it

0017 with 9. 5× speedup.

Performance and validation

On the Waymo Open Motion Dataset, FlowS (Ens.) achieves the highest Soft mAP (0. p. 5

Where the paper says it

BEST IN BOLD, SECOND UNDERLINED, THIRD MARKED ∗. FLOWS ACHIEVES THE HIGHEST MAP AND SOFT MAP WITH SINGLE-STEP INFERENCE (∼75 FPS).

4804) and mAP (0. p. 0
Where the paper says it

On the Waymo Open Motion Dataset, FlowS achieves state-of-the-art Soft mAP (0. 4804) and mAP (0.

p. 0
Where the paper says it

On the Waymo Open Motion Dataset, FlowS achieves state-of-the-art Soft mAP (0. 4804) and mAP (0.

4703), surpassing all published entries including ensemble methods. p. 5
Where the paper says it

4804) and mAP (0. 4703), surpassing all published entries including ensemble methods, while maintaining competitive displacement and miss rate.

The model runs at ∼75 FPS with single-step inference, demonstrating that local transport conditioning makes one-step generative motion prediction practical for safety-critical autonomy. p. 0
Where the paper says it

4804) and mAP (0. 4703) with ensemble at 75 FPS with singlestep inference, demonstrating that local transport conditioning makes one-step generative motion prediction practical for safetycritical autonomy.

Ablation studies show that adding the prior alone improves mAP by +0. p. 6
Where the paper says it

Without either, baseline Gaussian flow matching achieves 0. Adding the prior alone improves mAP by +0.

0082, and adding the step-consistent field alone yields +0. p. 6
Where the paper says it

0082 (shorter transport distance enables better mode capture). Adding the step-consistent field alone yields +0.

0056. The ensemble aggregates predictions from 6 independently trained models with varied hyperparameters (dropout rates, hidden dimensions, flow settings). p. 6
Where the paper says it

25 ms/scene, a 34% reduction over MTR. The ensemble aggregates predictions from 6 independently trained models with varied hyperparameters (dropout rates, hidden dimensions, flow settings).

Limitations and future work

The learned prior’s K = 6 mixture may not capture all taildistribution maneuvers (e.g., rare cyclist U-turns), especially true for cyclist maneuvers. p. 6

Where the paper says it

LIMITATIONS The learned prior’s K = 6 mixture may not capture all taildistribution maneuvers (e. , rare cyclist U-turns), especially true for cyclist maneuvers.

Our evaluation is limited to WOMD; transfer to datasets with different map representations, shorter horizons, or different agent densities would require encoder adaptation and prior recalibration. p. 6
Where the paper says it

While scaling K is a good solution, it increases memory linearly. Our evaluation is limited to WOMD; transfer to datasets with different map representations, shorter horizons, or different agent densities would require encoder adaptation and prior recalibration.

Extending FlowS to joint prediction via interaction-conditioned flows is a natural next step. p. 6
Where the paper says it

The current formulation predicts marginal per-agent futures without explicit joint consistency, which may produce overlapping trajectories in dense interactions (though overlap rate remains competitive; Tab. Extending FlowS to joint prediction via interaction-conditioned flows is a natural next step.

The paper concludes that anchoring the displacement field at learned priors along straight-line paths yields a stable, low-variance self-consistency target, reducing the training instability of prior formulations on curved diffusion paths. p. 7
Where the paper says it

58 displacement field that ensures one-step accuracy. Anchoring the displacement field at learned priors along straight-line paths yields a stable, low-variance self-consistency target, reducing the training instability of prior formulations on curved diffusion paths.

The final result is that FlowS reduces iterative refinement to a single deterministic forward pass at ∼75 FPS while achieving state-of-the-art mAP

--- Page 1 ---

FlowS: One-Step Motion Prediction via Local Transport Conditioning Leandro Di Bella, Bruno Cornelis, and Adrian Munteanu Member, IEEE Abstract—Generative motion prediction must satisfy three simultaneous requirements for real-world autonomy: high accuracy, diverse multimodal futures, and strictly bounded latency Diffusion models meet the first two but violate the third, requiring tens to hundreds of denoising steps. We identify a conditioning strategy that resolves this tension: single-step integration is accurate when the underlying transport problem is local. p. 0

Where the paper says it

Diffusion models meet the first two but violate the third, requiring tens to hundreds of denoising steps. We identify a conditioning strategy that resolves this tension: single-step integration is accurate when the underlying transport problem is local.

A model that must both discover the correct behavioral mode and traverse a long displacement in one step accumulates large discretization errors; conditioning the base distribution to lie near plausible futures reduces the problem to short-range refinement, the regime where a single Euler step suffices. p. 0
Where the paper says it

We identify a conditioning strategy that resolves this tension: single-step integration is accurate when the underlying transport problem is local. A model that must both discover the correct behavioral mode and traverse a long displacement in one step accumulates large discretization errors; conditioning the base distribution to lie near plausible futures reduces the problem to short-range refinement, the regime where a single Euler step suffices.

We instantiate this local transport conditioning in FlowS, a conditional flow matching framework with two mechanisms. p. 0
Where the paper says it

A model that must both discover the correct behavioral mode and traverse a long displacement in one step accumulates large discretization errors; conditioning the base distribution to lie near plausible futures reduces the problem to short-range refinement, the regime where a single Euler step suffices. We instantiate this local transport conditioning in FlowS, a conditional flow matching framework with two mechanisms.

First, an online, scene-conditioned learned prior emits K calibrated anchor trajectories per agent, each already near a plausible future, converting mode discovery into local correction. p. 0
Where the paper says it

We instantiate this local transport conditioning in FlowS, a conditional flow matching framework with two mechanisms. First, an online, scene-conditioned learned prior emits K calibrated anchor trajectories per agent, each already near a plausible future, converting mode discovery into local correction.

Second, a stepconsistent displacement field enforces semigroup self-consistency, guaranteeing that a single step inherits multi-step accuracy. p. 0
Where the paper says it

First, an online, scene-conditioned learned prior emits K calibrated anchor trajectories per agent, each already near a plausible future, converting mode discovery into local correction. Second, a stepconsistent displacement field enforces semigroup self-consistency, guaranteeing that a single step inherits multi-step accuracy.

Crucially, anchoring this field at learned priors along straightline paths yields a stable, low-variance training target, unlike prior self-consistency methods that suffer from high-variance bootstrap signals on curved diffusion paths. p. 0
Where the paper says it

Second, a stepconsistent displacement field enforces semigroup self-consistency, guaranteeing that a single step inherits multi-step accuracy. Crucially, anchoring this field at learned priors along straightline paths yields a stable, low-variance training target, unlike prior self-consistency methods that suffer from high-variance bootstrap signals on curved diffusion paths.

On the Waymo Open Motion Dataset, FlowS achieves state-of-the-art Soft mAP (0. p. 0
Where the paper says it

Crucially, anchoring this field at learned priors along straightline paths yields a stable, low-variance training target, unlike prior self-consistency methods that suffer from high-variance bootstrap signals on curved diffusion paths. On the Waymo Open Motion Dataset, FlowS achieves state-of-the-art Soft mAP (0.

4804) and mAP (0.4703) with ensemble at 75 FPS with singlestep inference, demonstrating that local transport conditioning makes one-step generative motion prediction practical for safetycritical autonomy. p. 0
Where the paper says it

4804) and mAP (0. 4703) with ensemble at 75 FPS with singlestep inference, demonstrating that local transport conditioning makes one-step generative motion prediction practical for safetycritical autonomy.

Code and pretrained models will be released upon acceptance

--- Page 2 ---

Fig. 1. FlowS architecture. p. 1

Where the paper says it

FlowS architecture. Scene inputs are processed by a Transformer-based encoder (Agent/Map).

Scene inputs are processed by a Transformer-based encoder (Agent/Map). p. 1
Where the paper says it

FlowS architecture. Scene inputs are processed by a Transformer-based encoder (Agent/Map).

The Learned Prior module replaces the standard Gaussian base by producing K scene-conditioned anchor trajectories. p. 1
Where the paper says it

Scene inputs are processed by a Transformer-based encoder (Agent/Map). The Learned Prior module replaces the standard Gaussian base by producing K scene-conditioned anchor trajectories.

During training, the decoder (Dθ) learns flow dynamics from anchors to ground-truth distributions via a step-consistent displacement loss (Lstep). p. 1
Where the paper says it

The Learned Prior module replaces the standard Gaussian base by producing K scene-conditioned anchor trajectories. During training, the decoder (Dθ) learns flow dynamics from anchors to ground-truth distributions via a step-consistent displacement loss (Lstep).

At inference, a single forward evaluation produces multi-agent trajectory predictions, filtered by NMS. p. 1
Where the paper says it

During training, the decoder (Dθ) learns flow dynamics from anchors to ground-truth distributions via a step-consistent displacement loss (Lstep). At inference, a single forward evaluation produces multi-agent trajectory predictions, filtered by NMS.

--- Page 3 ---

Fig. 2. Why a learned prior helps one-step flow matching. p. 2

Where the paper says it

Why a learned prior helps one-step flow matching. Left: Gaussian start. p. 2

Where the paper says it

Why a learned prior helps one-step flow matching. Left: Gaussian start.

Left: Gaussian start. A scene-agnostic base sample Z ∼ N (0, I) can lie far from the target future, so a single update must both identify the correct mode and traverse a long displacement, which increases error. p. 2
Where the paper says it

Left: Gaussian start. A scene-agnostic base sample Z ∼ N (0, I) can lie far from the target future, so a single update must both identify the correct mode and traverse a long displacement, which increases error.

Right: learned prior. p. 2
Where the paper says it

A scene-agnostic base sample Z ∼ N (0, I) can lie far from the target future, so a single update must both identify the correct mode and traverse a long displacement, which increases error. Right: learned prior.

Scene-conditioned anchors Ak already lie near plausible futures, so one-step CFM only needs a short corrective transport to reach the target. p. 2
Where the paper says it

Right: learned prior. Scene-conditioned anchors Ak already lie near plausible futures, so one-step CFM only needs a short corrective transport to reach the target.

The prior therefore improves conditioning by reducing transport length, yielding∥YGT − Ak∥ ≪∥YGT − Z∥2. p. 2
Where the paper says it

Scene-conditioned anchors Ak already lie near plausible futures, so one-step CFM only needs a short corrective transport to reach the target. The prior therefore improves conditioning by reducing transport length, yielding ∥YGT − Ak∥ ≪ ∥YGT − Z∥.

--- Page 4 ---

Fig. 3. Step-consistent displacement field. Left: the base CFM velocity field vϕ(Xt, t, c) along the transport path from anchor A to the target. p. 3

Where the paper says it

Step-consistent displacement field. Left: the base CFM velocity field vϕ(Xt, t, c) along the transport path from anchor A to the target.

Middle: semigroup consistency enforces that one 2d update matches two composed d updates. p. 3
Where the paper says it

Left: the base CFM velocity field vϕ(Xt, t, c) along the transport path from anchor A to the target. Middle: semigroup consistency enforces that one 2d update matches two composed d updates.

Right: at inference, the displacement field sψ(A, 0, 1, c) realizes the full transport in one pass, while preserving the infinitesimal limit limd→0 sψ = vϕ. p. 3
Where the paper says it

Middle: semigroup consistency enforces that one 2d update matches two composed d updates. Right: at inference, the displacement field sψ(A, 0, 1, c) realizes the full transport in one pass, while preserving the infinitesimal limit limd→0 sψ = vϕ.

--- Page 5 ---

Fig. 4. Prior–displacement field synergy. p. 4

Where the paper says it

The displacement field sψ addresses marginal curvature: as noted earlier, it absorbs the cumulative effect of curvature in vϕ via semigroup consistency, Fig. Prior–displacement field synergy.

Left: Minimum transport distance to ground truth for the learned prior (blue, mean 0. p. 4
Where the paper says it

Prior–displacement field synergy. Left: Minimum transport distance to ground truth for the learned prior (blue, mean 0.

78) vs. Gaussian N (0, I) (red, mean 1. p. 4
Where the paper says it

Left: Minimum transport distance to ground truth for the learned prior (blue, mean 0. Gaussian N (0, I) (red, mean 1.

23), showing a 36% reduction that enables accurate one-step integration. p. 4
Where the paper says it

Gaussian N (0, I) (red, mean 1. 23), showing a 36% reduction that enables accurate one-step integration.

Right: One-step velocity magnitude for prior+displacement field (blue, mean 1. p. 4
Where the paper says it

23), showing a 36% reduction that enables accurate one-step integration. Right: One-step velocity magnitude for prior+displacement field (blue, mean 1.

10) vs. displacement field only from N (0, I) (red, mean 1. p. 4
Where the paper says it

Right: One-step velocity magnitude for prior+displacement field (blue, mean 1. displacement field only from N (0, I) (red, mean 1.

72).

--- Page 6 ---

TABLE I WOMD TEST LEADERBOARD RESULTS BEST IN BOLD, SECOND UNDERLINED, THIRD MARKED ∗ FLOWS ACHIEVES THE HIGHEST MAP AND SOFT MAP WITH SINGLE-STEP INFERENCE (∼75 FPS) Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens.) [25] ICRA’25 0. p. 5

Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

4737 0.4531 0.5564∗ 1.1188 0.1084 0.1259∗ ModeSeq (Ens. p. 5
Where the paper says it

) [25] ICRA’25 0. 1259∗ ModeSeq (Ens.

p. 5
Where the paper says it

) [25] ICRA’25 0. 1259∗ ModeSeq (Ens.

) [26] CVPR’25 0. p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

4737 0.4665 0.5680 1.1766 0.1204 0.1275 BeTop [27] NeurIPS’24 0. p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

4698 0.4587 0.5716 1.1668 0.1183 0.1272 MGTR [28] ICRA’24 0. p. 5
Where the paper says it

1275 BeTop [27] NeurIPS’24 0. 1272 MGTR [28] ICRA’24 0.

4599, CVPR’25 0. p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

4599, CVPR’25 0.4505, ICRA’24) EDA [29] AAAI’24 0. p. 5
Where the paper says it

1272 MGTR [28] ICRA’24 0. 1275 EDA [29] AAAI’24 0.

3711, AAAI’24) 0. p. 5
Where the paper says it

1272 MGTR [28] ICRA’24 0. 1275 EDA [29] AAAI’24 0.

p. 5
Where the paper says it

1272 MGTR [28] ICRA’24 0. 1275 EDA [29] AAAI’24 0.

p. 5
Where the paper says it

1272 MGTR [28] ICRA’24 0. 1275 EDA [29] AAAI’24 0.

3711, AAAI’24) 0.3711, AAAI’24). p. 5
Where the paper says it

1272 MGTR [28] ICRA’24 0. 1275 EDA [29] AAAI’24 0.

MTR++ [15] TPAMI’24 0. p. 5
Where the paper says it

1275 EDA [29] AAAI’24 0. 1266 MTR++ [15] TPAMI’24 0.

4489, TPAMI’24) 0. p. 5
Where the paper says it

1275 EDA [29] AAAI’24 0. 1266 MTR++ [15] TPAMI’24 0.

4329, TPAMI’24) 0. p. 5
Where the paper says it

1275 EDA [29] AAAI’24 0. 1266 MTR++ [15] TPAMI’24 0.

p. 5
Where the paper says it

1275 EDA [29] AAAI’24 0. 1266 MTR++ [15] TPAMI’24 0.

5906, ICRA’25) 1. p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

1939 MTR [14] NeurIPS’22 TPAMI’24) 0. p. 5
Where the paper says it

1266 MTR++ [15] TPAMI’24 0. 1281 MTR [14] NeurIPS’22 0.

4403, NeurIPS’22) 0. p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

4329, TPAMI’24) 0.5964, NeurIPS’25) 1. p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

1888 HPTR [30] AAAI’24) 0. p. 5
Where the paper says it

1281 MTR [14] NeurIPS’22 0. 1274 HPTR [30] AAAI’24 0.

p. 5
Where the paper says it

1281 MTR [14] NeurIPS’22 0. 1274 HPTR [30] AAAI’24 0.

3968, AAAI’24) 0. p. 5
Where the paper says it

1272 MGTR [28] ICRA’24 0. 1275 EDA [29] AAAI’24 0.

3904, ICRA’24) p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

0.
p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

3968, ICRA’24). p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

--- Page 7 ---

TABLE II PER-CLASS PERFORMANCE ON THE VALIDATION SET OF WAYMO OPEN MOTION DATASET Setting Category mAP ↑ Vehicle 0.4756 Pedestrian 0. p. 0

Where the paper says it

The essential challenge is clear: given the recent scene with spatial positions and movement dynamics, predict where all agents will be within the next few seconds. A forecast delivered after the next control update is functionally useless: if a pedestrian steps off the curb, the braking decision cannot wait.

4808 Cyclist 0. p. 5
Where the paper says it

6707 Pedestrian 0. 3551 Cyclist 0.

3993 Avg 0. p. 5
Where the paper says it

3551 Cyclist 0. 6969 Avg 0.

4508, Avg 1. p. 6
Where the paper says it

Together, they achieve 0. 4508 mAP, the gains are complementary, not redundant.

1766

--- Page 8 ---

TABLE III COMPONENT ABLATION (SINGLE-STEP INFERENCE, VALIDATION SET) Prior Step-Consistent Field mAP ↑ minADE ↓ × × 0.4376 0.5916 × 0.4432 0.5850 × 0.4458 0.5823 0.4508, Avg, Avg

--- Page 9 ---

TABLE IV DRIFT ANALYSIS BY SCENE DENSITY MINADE/MINFDE (M) ON THE WOMD VALIDATION SET Sparse (1–3) 0.943 0.983 1.208 1.265 Medium (4–5) 0. p. 6

Where the paper says it

Sparse (1–3) 0. 265 532 Medium (4–5) 0.

843, Avg, Avg Sparse (1–3) 0. p. 5
Where the paper says it

3551 Cyclist 0. 6969 Avg 0.

943, Avg, Avg Dense (6+) 0. p. 5
Where the paper says it

3551 Cyclist 0. 6969 Avg 0.

734, Avg,Avg ∆ (Dense) −2. p. 5
Where the paper says it

3551 Cyclist 0. 6969 Avg 0.

99% −3.96%

--- Page 10 ---

TABLE V SCALE-AWARE DISPLACEMENT ANALYSIS DISPLACEMENT MAGNITUDE TARGET MAGNITUDE AND COSINE ALIGNMENT AT DIFFERENT POINTS ALONG THE FLOW PATH t d ∥sψ∥ ∥tgt∣cos(sψ, tgt) 0.00 1.00 0.729 0.730 16

--- Page 11 ---

TABLE VI PER-HORIZON ERROR ANALYSIS DISPLACEMENT ERROR (M) AT KEY PREDICTION HORIZONS ON THE WOMD VALIDATION SET Horizon Mean Median 95th Cum. ADE Horizon Mean Median 95th Cum. p. 7

Where the paper says it

DISPLACEMENT ERROR (M) AT KEY PREDICTION HORIZONS ON THE WOMD VALIDATION SET. Horizon Mean Median 95th 99th Cum.

p. 7
Where the paper says it

DISPLACEMENT ERROR (M) AT KEY PREDICTION HORIZONS ON THE WOMD VALIDATION SET. Horizon Mean Median 95th 99th Cum.

ADE

--- Page 12 ---

REFERENCES [3] Y. Lipman, R. T. Chen, H. Ben-Hamu, M. p. 7

Where the paper says it

16948, 2023. Ben-Hamu, M.

Nickel, and M. p. 7
Where the paper says it

Ben-Hamu, M. Nickel, and M.

Le “Flow matching for generative modeling” arXiv preprint arXiv:2210. p. 7
Where the paper says it

Nickel, and M. Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210.

02747 [4] C Jiang, A Cornman et al. p. 7
Where the paper says it

02747, 2022. Cornman, C.

, “Motiondiffuser: Controllable multi-agent motion prediction using diffusion” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. p. 7
Where the paper says it

Anguelov et al. , “Motiondiffuser: Controllable multi-agent motion prediction using diffusion,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp.

9644–9653 [11] A Tong, N Malkin et al. p. 7
Where the paper says it

, “Motiondiffuser: Controllable multi-agent motion prediction using diffusion,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 9644–9653.

, “Conditional flow matching: Simulation-free dynamic optimal transport” arXiv preprint arXiv:2302. p. 7
Where the paper says it

Wolf, and Y. Bengio, “Conditional flow matching: Simulation-free dynamic optimal transport,” arXiv preprint arXiv:2302.

p. 7
Where the paper says it

Wolf, and Y. Bengio, “Conditional flow matching: Simulation-free dynamic optimal transport,” arXiv preprint arXiv:2302.

p. 7
Where the paper says it

Wolf, and Y. Bengio, “Conditional flow matching: Simulation-free dynamic optimal transport,” arXiv preprint arXiv:2302.

p. 7
Where the paper says it

Wolf, and Y. Bengio, “Conditional flow matching: Simulation-free dynamic optimal transport,” arXiv preprint arXiv:2302.

00482 [8] B Liao, S Chen et al. p. 7
Where the paper says it

17 207–17 218. Chen, and Y.

p. 7
Where the paper says it

17 207–17 218. Chen, and Y.

p. 7
Where the paper says it

17 207–17 218. Chen, and Y.

, “Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2025, pp. p. 7
Where the paper says it

Zhang et al. , “Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving,” in Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp.

p. 7
Where the paper says it

Zhang et al. , “Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving,” in Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp.

p. 7
Where the paper says it

Zhang et al. , “Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving,” in Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp.

12 037–12 047

--- Page 13 ---

TABLE VI PER-HORIZON ERROR ANALYSIS DISPLACEMENT ERROR (M) AT KEY PREDICTION HORIZONS ON THE WOMD VALIDATION SET Horizon Mean Median 95th Cum. ADE Horizon Mean Median 95th Cum. ADE

--- Page 14 ---

REFERENCES [3] Y. Lipman, R. T Chen et al. p. 7

Where the paper says it

17 207–17 218. Chen, and Y.

p. 7
Where the paper says it

17 207–17 218. Chen, and Y.

p. 7
Where the paper says it

17 207–17 218. Chen, and Y.

p. 7
Where the paper says it

17 207–17 218. Chen, and Y.

p. 7
Where the paper says it

17 207–17 218. Chen, and Y.

, “Flow matching for generative modeling” arXiv preprint arXiv:2210. p. 7
Where the paper says it

Nickel, and M. Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210.

p. 7
Where the paper says it

Nickel, and M. Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210.

p. 7
Where the paper says it

Nickel, and M. Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210.

p. 7
Where the paper says it

Nickel, and M. Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210.

p. 7
Where the paper says it

Nickel, and M. Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210.

02747 [11] A Tong, N Malkin et al. p. 7
Where the paper says it

Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210. 02747, 2022.

p. 7
Where the paper says it

Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210. 02747, 2022.

p. 7
Where the paper says it

Le, “Flow matching for generative modeling,” arXiv preprint arXiv:2210. 02747, 2022.

, “Conditional flow matching: Simulation-free dynamic optimal transport” arXiv preprint arXiv:2302.00482 [8] B Liao, S Chen et al., “Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2025, pp. 12 037–12 047

--- Page 15 ---

TABLE V SCALE-AWARE DISPLACEMENT ANALYSIS DISPLACEMENT MAGNITUDE TARGET MAGNITUDE AND COSINE ALIGNMENT AT DIFFERENT POINTS ALONG THE FLOW PATH t d∥sψ∥∥tgt∣cos(sψ, tgt) 0.00 1.00 0.729 0.730

--- Page 16 ---

TABLE VI PER-HORIZON ERROR ANALYSIS DISPLACEMENT ERROR (M) AT KEY PREDICTION HORIZONS ON THE WOMD VALIDATION SET Horizon Mean Median 95th Cum. ADE

--- Page 17 ---

REFERENCES [3] Y. Lipman, R. T Chen et al., “Flow matching for generative modeling” arXiv preprint arXiv:2210.02747 [11] A Tong, N Malkin et al., “Conditional flow matching: Simulation-free dynamic optimal transport” arXiv preprint arXiv:2302.00482 [8] B Liao, S Chen et al., “Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2025, pp. 12 037–12 047

--- Page 18 ---

REFERENCES [3] Y. Lipman, R. T Chen et al., “Flow matching for generative modeling” arXiv preprint arXiv:2210.02747 [11] A Tong, N Malkin et al., “Conditional flow matching: Simulation-free dynamic optimal transport” arXiv preprint arXiv:2302.00482

--- Page 19 ---

REFERENCES [3] Y. Lipman, R. T Chen et al., “Flow matching for generative modeling” arXiv preprint arXiv:2210.02747

--- Page 26 ---

REFERENCES [3] Y. Lipman, R. T Chen et al., “Flow matching for generative modeling” arXiv preprint arXiv:2210.02747

--- Page 5 ---

Fig. 5. Quality–latency trade-off. p. 5

Where the paper says it

To assess training Fig. Quality–latency trade-off.

mAP (blue, left axis) vs. p. 5
Where the paper says it

Quality–latency trade-off. mAP (blue, left axis) vs.

latency (red, right axis) across 1–16 inference steps Fig. p. 5
Where the paper says it

mAP (blue, left axis) vs. latency (red, right axis) across 1–16 inference steps.

6 Qualitative results on WOMD. p. 5
Where the paper says it

5× speedup. Qualitative results on WOMD.

Dense intersections, yielding maneuvers, and branching scenarios

--- Page 6 ---

TABLE I WOMD TEST LEADERBOARD RESULTS BEST IN BOLD, SECOND UNDERLINED, THIRD MARKED ∗ FLOWS ACHIEVES THE HIGHEST MAP AND SOFT MAP WITH SINGLE-STEP INFERENCE (∼75 FPS) Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens.) [25] ICRA’25 0.4737 0.4531 0.5564∗ 1.1188 0.1084 0.1259∗ ModeSeq (Ens.) [26] CVPR’25 0.4737, CVPR’25) 0. p. 5

Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

4665, CVPR’25) 0. p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

5680, ICRA’24) 1. p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

1766 0.1204 0.1275 BeTop [27] NeurIPS’24) 0. p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

4698, NeurIPS’24) 0. p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

4587, NeurIPS’24) 0. p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

5716, ICRA’23) 1. p. 5
Where the paper says it

Method Venue Soft mAP ↑ mAP ↑ minADE ↓ minFDE ↓ Miss Rate ↓ Overlap Rate ↓ RMP-YOLO (Ens. ) [25] ICRA’25 0.

1668 0.1183 0.1272 MGTR [28] ICRA’24) 0. p. 5
Where the paper says it

1275 BeTop [27] NeurIPS’24 0. 1272 MGTR [28] ICRA’24 0.

4599, CVPR’25) 0. p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

4599, CVPR’25), ICRA’24) 0. p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

4505, CVPR’25), ICRA’24). p. 5
Where the paper says it

1259∗ ModeSeq (Ens. ) [26] CVPR’25 0.

EDA [29] AAAI’24) 0. p. 5
Where the paper says it

1272 MGTR [28] ICRA’24 0. 1275 EDA [29] AAAI’24 0.

3711, AAAI’24) 0.3711, AAAI'24). MTR++ [15] TPAMI’24) 0. p. 5
Where the paper says it

1275 EDA [29] AAAI’24 0. 1266 MTR++ [15] TPAMI’24 0.

4809, TPAMI’24), 0. p. 5
Where the paper says it

1275 EDA [29] AAAI’24 0. 1266 MTR++ [15] TPAMI’24 0.

4329, TPAMI'24), 0.5906, ICRA’25) 1.1939 MTR [14] NeurIPS’22) 0. p. 5
Where the paper says it

1266 MTR++ [15] TPAMI’24 0. 1281 MTR [14] NeurIPS’22 0.

4376, NeurIPS’25), 0. p. 5
Where the paper says it

) [26] CVPR’25 0. 1275 BeTop [27] NeurIPS’24 0.

4376, TPAMI'24), 0.5964, NeurIPS’25) 1.1888 HPTR [30] AAAI’24) 0.3968, AAAI'24), 0.3904, ICRA’24)

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved one-step generative motion prediction by conditioning it on local transport problems, which reduces discretization error when an agent must discover a correct behavioral mode and traverse a long displacement in one step. This allows for high accuracy at real-time speeds like 75 FPS using single-step inference.

  2. Enhanced scene context representation by using a symmetric MTR++ encoder that vectorizes agent histories and HD-map polylines into polyline tokens processed with query centric self attention in local coordinates, preserving locality and scaling to large neighborhoods. This results in per agent context features c that condition all subsequent modules effectively.

  3. Introduced a learned scene conditioned prior that emits K calibrated anchor trajectories per agent, which converts the mode discovery problem into local correction by ensuring anchors lie near plausible futures. This shortens transport distance, making the problem a short-range refinement, where a single Euler step suffices for accurate prediction.

  4. Implemented a step consistent displacement field spsi that enforces semigroup self consistency, guaranteeing that a single step inherits multi-step accuracy by ensuring two consecutive small steps must agree with one large step. This field absorbs the cumulative effect of marginal curvature, stabilizing the bootstrap signal during training and ensuring the d=1 limit matches the flow model vphi.

  5. Developed a dual mechanism where anchors suppress transport distance and spsi absorbs marginal curvature to address independent sources of single-step error. The prior reduces mean transport distance by 36%, while the displacement field absorbs the cumulative effect of curvature in vphi via semigroup consistency.

Abstract

Generative motion prediction must satisfy three simultaneous requirements for real-world autonomy: high accuracy, diverse multimodal futures, and strictly bounded latency. Diffusion models meet the first two but violate the third, requiring tens to hundreds of denoising steps. We identify a conditioning strategy that resolves this tension: single-step integration is accurate when the underlying transport problem is local. A model that must both discover the correct behavioral mode and traverse a long displacement in one step accumulates large discretization errors; conditioning the base distribution to lie near plausible futures reduces the problem to short-range refinement, the regime where a single Euler step suffices. We instantiate this local transport conditioning in FlowS, a conditional flow matching framework with two mechanisms. First, an online, scene-conditioned learned prior emits K calibrated anchor trajectories per agent, each already near a plausible future, converting mode discovery into local correction. Second, a step-consistent displacement field enforces semigroup self-consistency, guaranteeing that a single step inherits multi-step accuracy. Crucially, anchoring this field at learned priors along straight-line paths yields a stable, low-variance training target, unlike prior self-consistency methods that suffer from high-variance bootstrap signals on curved diffusion paths. On the Waymo Open Motion Dataset, FlowS achieves state-of-the-art Soft mAP (0.4804) and mAP (0.4703) with ensemble at 75,FPS with single-step inference, demonstrating that local transport conditioning makes one-step generative motion prediction practical for safety-critical autonomy. Code and pretrained models will be released upon acceptance.

Sources

Related papers