Destination Support Restoration for Finite-Set Multimodal Trajectory Prediction

Fengrui Liu; Jiajun Peng; Duo Peng; Feng Liu

cs.ROSep 22, 2026

Destination Support Restoration for Finite-Set Multimodal Trajectory Prediction

Authors: Fengrui Liu, Jiajun Peng, Duo Peng, Feng Liu

Organizations: School of Computer Science and Technology, East China Normal University, Shanghai, China · School of Data Science, University of Science and Technology of China, Hefei, China · School of Computer Science and Technology, Tongji University, Shanghai, China · School of Psychology, Shanghai Jiao Tong University, Shanghai, China

Abstract

Robots operating around pedestrians often reason over a finite set of predicted human futures. Repeated online updates can concentrate this limited prediction budget on dominant destinations and leave plausible alternatives underrepresented or absent, removing those alternatives from the finite representation available to downstream decision making. We introduce Destination Support Restoration (DSR), a causal post-selection operator that repairs destination support without retraining the host predictor or increasing the maintained set size. At a repair step, DSR evaluates a temporary destination-stratified candidate bank from the observed prefix, converts candidate evidence into integer target counts, protects representatives of active modes, and reallocates redundant surplus hypotheses to deficient modes. The maintained and returned sets retain exactly $N$ hypotheses, and DSR replaces at most $\lceilρN\rceil$ entries. Protected representatives preserve current categorical support; lineage-aware particle filters also preserve surviving resampling ancestors. Each replacement reduces the allocation mismatch to the evidence-driven target by one. On the complete 3,719-trajectory Edinburgh protocol over three seeds, DSR reduces MIF weighted ADE and FDE by 13.36% and 13.30% at $N=64$ . Paired integrations with CLiFF, PPT, causal GDTS, Social Informer, and PECNet improve both metrics in every evaluated pair. These results show that finite-set support allocation is a useful prediction-side control point when a fixed hypothesis set serves as the interface to downstream systems.

Figures & tables

Fig. 1: Overview of Destination Support Restoration (DSR). (a) An observed prefix remains compatible with multiple destination modes. (b) A fixed-size maintained set concentrates on dominant modes. (c) Resampling may eliminate a still-plausible destination from the finite representation. (d) DSR protects surviving support and reallocates surplus slots toward evidence-supported deficits while returning the same number of hypotheses. (e) The figure illustrates the lineage-aware case; set-based hosts use protected mode representatives instead.

Fig. 2: Host update and DSR pipeline. The host first propagates and weights its maintained hypotheses and then checks the effective sample size. When $\mathrm{ESS}<\eta N$ , DSR maps evidence from a temporary destination-stratified candidate bank to target counts, protects active-mode representatives, replaces only unprotected surplus hypotheses, and returns $N$ equally weighted hypotheses to the host.

Predictor	Configuration	ADE/AOE $\downarrow$	FDE/FOE $\downarrow$	NLL $\downarrow$	DSR Reduction
Social Force [ 16 , 26 ] †	-	3.1240	3.9090	-	-
LSTM [ 16 , 27 ] †	-	2.1320	3.0050	-	-
Social LSTM [ 1 , 16 ] †	-	1.5240	2.5100	-	-
Attention LSTM [ 16 , 28 ] †	-	0.9860	1.3110	-	-
Social GAN [ 2 , 16 ] †	340 outputs	1.0420	2.0880	-	-
MIF [ 16 ]	$N=64$	0.6594	1.2420	5.1041	-

TABLE I: Results on Edinburgh. Lower is better. We include historical AOE/FOE values marked with

\dagger

as context because they follow a different protocol. NLL comparisons are paired within host; for CLiFF, we measure the DSR reduction relative to GoalBridge.

Host	Scene	wADE $\downarrow$	wFDE $\downarrow$
MIF	ETH	$1.2456\to\mathbf{1.2255}$	$2.6604\to\mathbf{2.5983}$
	HOTEL	$0.4133\to\mathbf{0.4066}$	$0.9036\to\mathbf{0.8825}$
	UNIV	$0.7292\to\mathbf{0.7277}$	$1.5781\to\mathbf{1.5699}$
	ZARA1	$0.5258\to\mathbf{0.5185}$	$1.1513\to\mathbf{1.1306}$
	ZARA2	$0.4222\to\mathbf{0.4171}$	$0.9280\to\mathbf{0.9134}$
CLiFF	ETH	$1.1324\to\mathbf{1.1205}$	$2.2120\to\mathbf{2.1076}$

TABLE II: Paired ETH/UCY results over three seeds. Each cell reports Host

\to

DSR; lower is better. UNIV denotes stu03 for MIF and CLiFF.

Variant	wADE $\downarrow$	wFDE $\downarrow$
Full DSR	0.5713	1.0768
Zero repair ( $\rho=0$ )	0.69391	1.30667
w/o lineage + surplus protection	0.58088	1.09413
Random donor	0.59654	1.09956
Random candidate	0.58372	1.09875

TABLE III: Component ablation on MIF with

N=64

. Lower is better.

Mechanism	Random Reset	DSR	Gain (pp)
Immediate revival	2.94%	24.11%	21.17
Historical recovery	3.15%	22.21%	19.07
Pre-extinction increase	1.98%	22.72%	20.74

TABLE IV: Support-repair rates on MIF with

N=64

. Gain is the cluster-paired difference between method-specific rates in percentage points (pp).

Explore similar work

Harnessing Pre-Resolution Signals for Future Prediction Agents
Apr 17, 2026Chuyang Wei, Maohang Gao, Zhixin Han +12
University of Science and Technology of China · Zhongguancun Academy, Beijing, China · IIIS, Tsinghua University
From Explanations to Interventions: Execution-Guided Counterfactual Synthesis in Temporal Graphs
Sep 10, 2026Minwoo Yu, Young-guk HaCounterfactual Explanation Temporal Knowledge Graphs
Smart Computing Laboratory, Department of Computer Science & Engineering, Konkuk University, Seoul 05029, Republic of Korea
Agent-Centric Social Trajectory Prediction: A Free Energy Principle Perspective
May 25, 2026Yanping Wu, Ji Zhang, Hao Chen +2Ego-Trajectory Prediction Spatial Encoding
University of Glasgow Glasgow, UK · Southwest Jiaotong University Chengdu, China

Apr 17, 2026, cs.AIJ/K move · Enter open · S save

Harnessing Pre-Resolution Signals for Future Prediction Agents

Chuyang Wei, Maohang Gao, Zhixin Han, Kefei Chen+11

University of Science and Technology of China · Zhongguancun Academy, Beijing, China · IIIS, Tsinghua University

Open paper AskPDF arXiv

Many high-stakes decisions depend on forecasts made before outcomes are known. In this future prediction setting, the central challenge is that public evidence evolves over time, while the main supervision signal arrives only after resolution: the realized outcome mainly assesses final correctness, offering only coarse guidance on what to track, what to verify, and which judgments to leave uncertain along the way. Our key observation is that revisiting the same unresolved question over time creates informative temporal contrasts across evolving evidence and repeated forecasts, exposing what earlier attempts missed before resolution and yielding a diagnostic signal we call the pre-resolution signal. We instantiate this idea in Milkyway, a future prediction agent with a persistent future prediction harness, an editable external state that stores reusable procedural guidance across revisits to the same unresolved question. As the same unresolved question is revisited, Milkyway extracts pre-resolution signals from evolving evidence and repeated forecasts, uses them to update the harness, and improves later forecasts on that question before resolution. After resolution, the realized outcome serves as a post-resolution check of provisional updates. On the FutureX and FutureWorld benchmarks, Milkyway achieves strong performance against competitive baselines, and a mechanism study suggests that the gains stem from harness evolution driven by pre-resolution signals rather than repeated prediction alone.

Abstract

Explore similar work

Harnessing Pre-Resolution Signals for Future Prediction Agents

From Explanations to Interventions: Execution-Guided Counterfactual Synthesis in Temporal Graphs

Agent-Centric Social Trajectory Prediction: A Free Energy Principle Perspective