FloCOS: Predicting Compact Forward Occupancy Sets of Surrounding Vehicles via Conditional Flow Matching
Authors: Chuansheng Dong, Yinxiao Zhan, Zongdian Li, Peng Hou, Jin Wang, Jiayu Fan
Organizations: Chuansheng Dong, Peng Hou, Jin Wang, and Jiayu Fan are with the School of Future Science and Engineering, Soochow University, Suzhou 215222, China · Yinxiao Zhan is with the College of Control Science and Engineering, Zhejiang University, Hangzhou 310027, China · Zongdian Li is with the State Key Laboratory of Blockchain and Data Security, Zhejiang University, Hangzhou 310007, China
Existing methods for predicting surrounding vehicles' forward occupancy may under-approximate the ground-truth occupancy, exposing the ego vehicle to collision risk, or produce overly conservative regions that unnecessarily restrict feasible space and cause planning failures. To address these issues, we propose FloCOS, a conditional flow-matching framework that predicts compact forward occupancy sets with high occupancy coverage. FloCOS represents the forward occupancy set as a time-varying union of half-space convex polygons parameterized by temporal basis coefficients. Conditional flow matching captures the conditional distribution of these coefficients to govern the temporal evolution of the convex components, trained under coverage and union-area losses that penalize missed occupancy and excess area. For downstream trajectory optimization, the predicted nonconvex union is directly incorporated into hyperplane-based collision-avoidance constraints. Evaluations on the Waymo Open Motion Dataset show that FloCOS generates compact forward occupancy sets with high coverage and achieves a higher interaction completion rate than evaluated baselines. Real-world vehicle experiments further confirm its ability to support collision-free planning in interactive scenarios.
Figures & tables
Fig. 1: Effects of occupancy prediction on planning. All three panels use the same observation history and ego state. Left: undercoverage leaves part of the surrounding vehicle’s ground-truth occupancy uncovered, exposing the ego plan to collision risk. Middle: an overly conservative forward occupancy set blocks an otherwise feasible passage. Right: FloCOS covers the ground-truth occupancy in this scene and preserves space for the ego vehicle to pass.
Fig. 2: FloCOS generates temporal geometry coefficients from vehicle histories and map features and decodes them into sequences of convex components. The rightmost column shows the union of all eight components at selected physical prediction times; intermediate nodes and some individual components are omitted for clarity. Here, flow time τ∈[0,1] parameterizes the ODE integration path.
Fig. 3: Representative planning outcomes in three simulated interactions. Each column shows the same scene and the same planner, with the prediction method varying across rows; rows correspond to MTR-GMM + FORCE-OPT, Traffic-Rule Set Prediction, and FloCOS. Blue and orange–red denote the EV and SV, respectively, with light and dark traces indicating history and future motion. Green and purple dashed boundaries show the predicted SV occupancy at 1s and 2s , respectively.
Methods
Coverage (%, 1s / 2s )
Area ratio ( 1s / 2s )
Completion (%) [95% CI]
MTR [ 1 ]
88.5 / 88.2
2.15 / 3.76
47.4 [44.3, 50.5]
DenseTNT [ 29 ]
91.8 / 91.4
2.48 / 4.88
23.0 [20.5, 25.7]
M2I [ 30 ]
92.8 / 92.5
2.58 / 5.16
18.9 [16.6, 21.4]
TrajFlow [ 3 ]
92.1 / 91.6
2.20 / 3.62
48.8 [45.7, 51.9]
Traffic-Rule Set Prediction [ 20 ]
86.4 / 83.3
5.04 / 8.59
3.1 [2.2, 4.4]
MTR-GMM + FORCE-OPT [ 1 , 22 ]
72.8 / 69.7
1.61 / 2.18
85.2 [82.9, 87.3]
TABLE I: Prediction and Planning Results on 1000 Independent WOMD Test Scenarios
Configuration
Coverage (%, 1s / 2s )
Area ratio ( 1s / 2s )
Completion (%)
FloCOS
99.7 / 99.3
1.70 / 3.10
85.6
Traj-CFM
99.4 / 66.1
2.91 / 3.84
9.6
Deterministic
56.1 / 21.3
2.64 / 2.77
64.8
Gaussian
17.7 / 13.4
1.73 / 2.99
61.1
Component-Sum Area
26.6 / 26.0
2.46 / 4.62
57.2
Single-Component Coverage
99.4 / 99.1
2.03 / 3.99
78.1
TABLE II: Ablation results for FloCOS design choices.
Fig. 4: Real-world interaction experiments with FloCOS. (a)–(f) Head-on encounter; (g)–(i) lateral cut-in. Panels (j) and (k) respectively present the EV motion data in the head-on and lateral cut-in scenarios. In each panel, the upper plot shows EV speed and the lower plot shows EV heading; solid blue curves correspond to the real-world experiments with FloCOS, and dashed magenta curves show matched simulations without FloCOS. Crosses mark EV–SV collisions in the simulations.
Methods
Parameters ( 106 )
FLOPs ( 109 /window)
MTR family † [ 1 ]
64.150
13.925
DenseTNT [ 29 ]
31.115
91.495
M2I [ 30 ]
93.567
184.193
TrajFlow [ 3 ]
51.587
12.932
DOS-Ellipse [ 17 ]
64.341
13.933
FloCOS
0.372
0.032
TABLE III: Prediction Network Size and Computation
Safe navigation in dynamic environments requires anticipating future environmental states to account for spatiotemporal risks, specifically when and where collisions may occur. To this end, occupancy grid map (OGM) prediction has been widely adopted as an effective approach. However, existing OGM-based navigation methods often struggle to achieve accurate and efficient forecasting and fail to fully exploit the temporal information in predicted OGMs during planning. To address these challenges, we propose FORTE, a navigation framework that directly exploits the spatiotemporal evolution of predicted occupancy from the perspectives of spatiotemporal occupancy overlap and occupancy directivity. Based on these properties, FORTE evaluates multiple topology-distinct paths and selects the suitable one without explicit object detection or tracking. To support online planning, we formulate a latent diffusion model-based OGM predictor that generates the entire forecast horizon in a non-autoregressive manner while maintaining temporal consistency through temporal shift modules. Extensive evaluations demonstrate that FORTE outperforms state-of-the-art baselines. For prediction, FORTE achieves up to 215.3% higher IoU and 5.24x faster inference; for navigation, it yields up to a 3.5x higher success rate.
Hahjin Lee, Young J. Kim
Department of Computer Science and Engineering at Ewha Womans University in Korea
Roadside 4D occupancy forecasting predicts future semantic occupancy in a fixed infrastructure coordinate system. Roadside scenes are globally structured yet locally evolving, requiring comprehensive scene-level context while precisely modeling sparse localized motion. However, existing forecasting paradigms struggle to tackle both requirements. To address this limitation, we propose InfraWorld, a world model formulating roadside occupancy forecasting as selective state evolution. InfraWorld maintains a dense BEV state for the complete scene and persistent motion queries for moving regions, assigning complementary states to global context and local dynamics. Specifically, we propose Dynamic Query Initiation (DQI) to initialize queries in regions supported by motion in the observed history. During rollout, the dense state is updated through Long--Short Temporal Fusion (LST), which combines short-term adjacent-frame continuity with long-term motion trends to preserve stable context while capturing scene evolution. Motion-Aware Query Injection (MAQI) then refines the queries and injects confidence-weighted query information into corresponding BEV tokens, enabling targeted dynamic updates without perturbing global context. Together, InfraWorld concentrates predictive capacity on evolving traffic regions. Extensive experiments on the InfraOcc and Occ3D--nus datasets demonstrate that InfraWorld achieves state-of-the-art performance for roadside 4D occupancy forecasting while retaining its effectiveness under ego-centric settings. Code will be released.
Songkai Wang, Xiaokai Bai, Lei Yang +1
College of Information Science and Electronic Engineering, Zhejiang University. · School of Mechanical and Aerospace Engineering, Nanyang Technological University.
Legged robots under sparse waypoint guidance must avoid moving obstacles using partial, rapidly changing LiDAR observations. We present LOOP (Latent-recurrent Occupancy rollOut Policy), a local avoidance policy that connects sparse waypoint guidance to a frozen locomotion controller at 50 Hz. From occupancy and ego-velocity histories, a recurrent predictor forecasts future occupancy over a 1 s horizon by warping the current map with learned flow and visibility gates. These maps guide velocity selection through map-derived features and geometric risk estimates, providing an explicit interface for inspecting and replacing predictions. In encounter-synchronised Isaac Lab evaluations, LOOP achieves 57.1% head-on success at obstacle speeds of 2.5-3.2 m/s, exceeding a retrained reactive baseline by 8.2 percentage points. Comparisons with a rollout-free BEV policy show smaller, scenario-dependent gains from the prediction branch, including improved crossing success and reduced variability across training seeds at the highest head-on speeds. The adapter runs onboard a Unitree Go2 in 14.5 ms per step and completes all 16 real-world crossing trials without collision, demonstrating deployment feasibility.
Yuhui Mao, Fen Liu, Shenghai Yuan +3
School of Electrical and Electronic Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798