CATP: Context-Aware Trajectory Prediction with Competition Symbiosis

TL;DR

This work tackles trajectory prediction under rich, dynamic context by introducing CATP, a context-aware model built on a manager–worker framework. The manager selects the best worker for a given context, while workers specialize on context-specific moving patterns; training follows a competition symbiosis paradigm with a data-driven, regularized update rule and a top-$k$ ADE objective. Empirical results demonstrate CATP outperforms state-of-the-art baselines on context-rich datasets (e.g., DOTA, Bird-8) and generalizes to context-aware time-series tasks, with ablations clarifying training dynamics and robust settings. The approach offers a scalable, modular path to leveraging context in sequential predictions, with practical implications for navigation, wildlife tracking, and competitive multi-agent systems.

Abstract

Contextual information is vital for accurate trajectory prediction. For instance, the intricate flying behavior of migratory birds hinges on their analysis of environmental cues such as wind direction and air pressure. However, the diverse and dynamic nature of contextual information renders it an arduous task for AI models to comprehend its impact on trajectories and consequently predict them accurately. To address this issue, we propose a ``manager-worker'' framework to unleash the full potential of contextual information and construct CATP model, an implementation of the framework for Context-Aware Trajectory Prediction. The framework comprises a manager model, several worker models, and a tailored training mechanism inspired by competition symbiosis in nature. Taking CATP as an example, each worker needs to compete against others for training data and develop an advantage in predicting specific moving patterns. The manager learns the workers' performance in different contexts and selects the best one in the given context to predict trajectories, enabling CATP as a whole to operate in a symbiotic manner. We conducted two comparative experiments and an ablation study to quantitatively evaluate the proposed framework and CATP model. The results showed that CATP could outperform SOTA models, and the framework could be generalized to different context-aware tasks.

Figures (3)

• Figure 1: A data sample consists of three parts, all in matrix form. The shapes of $C$, $TX$, and $TY$ are $LC \times (N + \sum(LE^i - 1))$, $LX \times 2|U|$, and $LY \times 2$, respectively.
• Figure 2: CATP consists of a manager model (layers L1-L4) and multiple worker models (layers L5-L9). The manager selects the optimal worker in a given context $C$, and the selected worker predicts $\hat{TY}$ based on $TX$ accurately.
• Figure 3: The proposed training mechanism. We first train workers based on the manager's results and then train the manager based on workers' performance.