🤖 AI Summary
To address mode collapse, low generation quality, insufficient multimodal information utilization, and high inference latency in autonomous driving–oriented pedestrian trajectory forecasting, this paper proposes Goal-Guided Diffusion for Trajectory Synthesis (GDTS). GDTS is built upon denoising diffusion probabilistic models (DDPMs), integrating feature sharing and progressive denoising. Its core contributions are: (1) a goal-estimation prior that enables explicit modeling of pedestrian motion intent via a target-conditioned encoding mechanism; and (2) a two-stage hierarchical sampling algorithm that generates diverse trajectories in parallel within a single forward pass, balancing modality richness and inference efficiency. Evaluated on mainstream public benchmarks, GDTS achieves state-of-the-art performance across accuracy, diversity, and computational efficiency—delivering near real-time inference at ≈30 FPS while simultaneously improving all three metrics.
📝 Abstract
Accurate prediction of pedestrian trajectories is crucial for improving the safety of autonomous driving. However, this task is generally nontrivial due to the inherent stochasticity of human motion, which naturally requires the predictor to generate multi-modal prediction. Previous works leverage various generative methods, such as GAN and VAE, for pedestrian trajectory prediction. Nevertheless, these methods may suffer from mode collapse and relatively low-quality results. The denoising diffusion probabilistic model (DDPM) has recently been applied to trajectory prediction due to its simple training process and powerful reconstruction ability. However, current diffusion-based methods do not fully utilize input information and usually require many denoising iterations that lead to a long inference time or an additional network for initialization. To address these challenges and facilitate the use of diffusion models in multi-modal trajectory prediction, we propose GDTS, a novel Goal-Guided Diffusion Model with Tree Sampling for multi-modal trajectory prediction. Considering the"goal-driven"characteristics of human motion, GDTS leverages goal estimation to guide the generation of the diffusion network. A two-stage tree sampling algorithm is presented, which leverages common features to reduce the inference time and improve accuracy for multi-modal prediction. Experimental results demonstrate that our proposed framework achieves comparable state-of-the-art performance with real-time inference speed in public datasets.
