AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

RAIN: Reinforced Hybrid Attention Inference Network for Motion Forecasting

Jiachen Li, Fan Yang, Hengbo Ma, Srikanth Malla, Masayoshi Tomizuka, Chiho Choi

ICCV 2021 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Interactive multi-agent trajectory prediction · 42 citations (OpenAlex)

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 changedWhereMeasuredEffect
add RL-based hard attention (hybrid) instead of soft attention only (ablation)RAIN · nuScenes vehiclesmiss rate @2.0m (%): 21.0 -> 11.3Improved
add RL-based hard attention (hybrid) instead of soft attention only (ablation)RAIN · nuScenes pedestriansmiss rate @1.0m (%): 13.2 -> 11.7Improved
use RL-based hard attention instead of ELBO-based module (ablation)RAIN · nuScenes vehiclesmiss rate @2.0m (%): 16.1 -> 11.3Improved
use RL-based hard attention instead of ELBO-based module (ablation)RAIN · Mixed particle system relation recognitionaccuracy (%): 89.31 -> 93.52Improved
use dynamic (recurrent) hard attention instead of static (ablation)RAIN · nuScenes vehiclesmiss rate @2.0m (%): 11.3 -> 9.5Improved
use dynamic (recurrent) hard attention instead of static (ablation)RAIN · nuScenes pedestriansmiss rate @1.0m (%): 11.7 -> 9.8Improved
use dynamic (recurrent) hard attention instead of static (ablation)RAIN · nuScenes vehiclesminFDE20 at 4.0s (m): 1.17 -> 1.12Improved
add vehicle kinematics model to motion generatorRAIN (full+soft) · nuScenes vehiclesminFDE20 at 4.0s (m): 1.27 -> 1.28No clear effect
add hard attention over history soft attention only (HisRepItself)RAIN · Human3.6M WalkingMAE 1k ms: 0.64 -> 0.60Improved
add hard attention over history soft attention only (HisRepItself)RAIN · Human3.6M DiscussionMAE 400 ms: 0.87 -> 0.91Hurt
use RAIN hybrid attention instead of DNRIRAIN · Mixed particle system relation recognitionaccuracy (%): 87.54 -> 93.52Improved
use RAIN instead of fully supervised relation labelsRAIN · Mixed particle system relation recognitionaccuracy (%): 97.48 -> 93.52Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy