Research map / Multi-agent systems and games / Human-AI and goal-conditioned
SHARP: Short-Window Streaming for Accurate and Robust Prediction in Motion Forecasting
Alexander Prutsch, Christian Fruhwirth-Reisinger, David Schinagl, Horst Possegger
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add target-centric features to baseline streaming (ablation) | baseline streaming · AV2 val (Tcl=5s) | b-mFDE6: 1.91 -> 1.84 | Improved |
| add instance-aware context streaming (ablation) | SHARP · AV2 val (Tcl=5s) | b-mFDE6: 1.84 -> 1.81 | Improved |
| add dual training at long context (ablation) | SHARP · AV2 val (Tcl=5s) | b-mFDE6: 1.81 -> 1.82 | No clear effect |
| add target-centric features (ablation) | baseline streaming · AV2 val (Tcl=1s) | b-mFDE6: 3.28 -> 3.18 | Improved |
| add instance-aware streaming without dual training (ablation) | SHARP · AV2 val (Tcl=1s) | b-mFDE6: 3.18 -> 3.25 | Hurt |
| add dual training at short context (ablation) | SHARP · AV2 val (Tcl=1s) | b-mFDE6: 3.25 -> 2.13 | Improved |
| add instance-aware streaming (ablation) | SHARP · AV1 val (Tcl=2s) | b-mFDE6: 1.58 -> 1.52 | Improved |
| add dual training at short context (ablation) | SHARP · AV1 val (Tcl=0.5s) | b-mFDE6: 1.78 -> 1.61 | Improved |
| replace implicit context relay with instance-aware streamer at long context (ablation) | SHARP · AV2 val (long input) | b-mFDE6: 1.33 -> 1.08 | Improved |
| train with 1s windows over 8 passes plus instance-aware streaming (vs 3s baseline setup) (ablation) | SHARP · AV2 val (Tcl=5s) | b-mFDE6: 1.84 -> 1.81 | Improved |
| weight chunk loss twice as much as stream loss (ablation) | SHARP · AV2 val (Tcl=1s vs 5s) | b-mFDE6: 2.70 -> 2.52 | Mixed |
| add joint refinement module over naive combination of marginals (ablation) | SHARP · AV2 multi-agent val | potential collision events: 5,533 -> 973 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DeMo — DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024) · better on 8 of 12 shared results
- SRefiner — SRefiner: Soft-Braid Attention for Multi-Agent Trajectory Refinement (ICCV 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Motion Forecasting in Continuous Driving (NeurIPS 2024 · similarity 0.90)
- Query-Centric Trajectory Prediction (CVPR 2023 · similarity 0.87)
- MESSENGER: Memory-Enhanced Sequential Scene Flow Estimation via Autoregressive Next-Frame Forecasting (NeurIPS 2026 · similarity 0.87)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.87)
- SEPT: Towards Efficient Scene Representation Learning for Motion Prediction (ICLR 2024 · similarity 0.86)
- SmartPretrain: Model-Agnostic and Dataset-Agnostic Representation Learning for Motion Prediction (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
