THOMAS:学習型マルチエージェントサンプリングによる軌道ヒートマップ出力
THOMAS: Trajectory Heatmap Output with learned Multi-Agent Sampling
複数エージェントの将来軌道をヒートマップとして推定し、学習可能な再結合モデルでモダリティ間の整合性を取る統合予測フレームワークを提案。Interactionチャレンジで1位を獲得。
著者: Thomas Gilles, Stefano Sabatini, Dzmitry Tsishkou, Bogdan Stanciulescu, Fabien Moutarde
分類: cs.CV, cs.RO
原文アブストラクト
In this paper, we propose THOMAS, a joint multi-agent trajectory prediction framework allowing for an efficient and consistent prediction of multi-agent multi-modal trajectories. We present a unified model architecture for simultaneous agent future heatmap estimation, in which we leverage hierarchical and sparse image generation for fast and memory-efficient inference. We propose a learnable trajectory recombination model that takes as input a set of predicted trajectories for each agent and outputs its consistent reordered recombination. This recombination module is able to realign the initially independent modalities so that they do no collide and are coherent with each other. We report our results on the Interaction multi-agent prediction challenge and rank $1^{st}$ on the online test leaderboard.
