src.model.model module

src.model.model module#

class src.model.model.Model(*args: Any, **kwargs: Any)[source]#

Bases: Module

Pedestrian trajectory prediction model based on Transformer and diffusion.

The model combines pedestrian history, social interaction with nearby pedestrians, interaction with surrounding vehicles, and static map context to predict motion intent, either acceleration or noise, during reverse denoising in DDPM/DDIM.

__init__(args)[source]#

Initialize model layers and embedding modules.

Parameters:

args (Namespace) – Configuration object containing: - model_dim (int): Internal model feature dimension. - map_feature_dim (int): Intermediate map feature dimension. - lstm_layer_num (int): Number of LSTM layers for temporal data. - head_num (int): Number of attention heads. - attention_layer_num (int): Number of Transformer decoder layers. - latent_token_num (int): Number of latent tokens used to compress map features. - dropout (float): Dropout ratio. - pred_step (int): Prediction horizon. - use_spatial_anchor (bool): Whether to enhance map positional encoding with spatial anchors.

set_ped_embedding(pos: torch.FloatTensor, vel: torch.FloatTensor, hst: torch.FloatTensor, des: torch.FloatTensor, spd: torch.FloatTensor)[source]#

Compute and store the joint pedestrian embedding.

This method maps pedestrian position, velocity, history, destination, and desired speed into a high-dimensional space and sums them to form the initial pedestrian feature vector. It also computes positional encoding.

Parameters:

  • pos (torch.FloatTensor) – Current pedestrian coordinates (x, y). Shape: (batch_size, num_peds, 2)

  • vel (torch.FloatTensor) – Current pedestrian velocity (vx, vy). Shape: (batch_size, num_peds, 2)

  • hst (torch.FloatTensor) – Pedestrian history trajectory sequence. Shape: (batch_size, num_peds, hist_step, 2)

  • des (torch.FloatTensor) – Pedestrian destination coordinates. Shape: (batch_size, num_peds, 2)

  • spd (torch.FloatTensor) – Pedestrian desired speed scalar. Shape: (batch_size, num_peds, 1)

Side Effects:

Sets self.ped_embedding: fused pedestrian features. Sets self.pos: cached current positions for later map indexing. Sets self.pe: positional encoding features.

set_veh_embedding(veh: torch.FloatTensor)[source]#

Compute and store vehicle feature embeddings.

Processes vehicle trajectory history with an LSTM. If the current scene contains no vehicles, NaN padding is inserted automatically.

Parameters:

veh (torch.FloatTensor) – Vehicle history trajectory sequence. Shape: (batch_size, num_vehs, hist_step + 1, 2)

Side Effects:

Sets self.veh_embedding: vehicle feature vectors.

set_map_embedding(map: torch.FloatTensor, xmin: torch.FloatTensor, xmax: torch.FloatTensor, ymin: torch.FloatTensor, ymax: torch.FloatTensor)[source]#

Compute and store static map embeddings.

A CNN extracts local features from the rasterized map and combines them with absolute positional encoding. To reduce computation, latent queries compress the dense map features through cross-attention into a compact environmental context representation.

Parameters:

  • map (torch.FloatTensor) – Rasterized environment map, where 0 denotes walkable area and 1 denotes obstacles. Shape: (Map_W, Map_H), with the first dimension as x pointing right and the second as y pointing up.

  • xmin (float) – Minimum x value of the map in world coordinates.

  • xmax (float) – Maximum x value of the map in world coordinates.

  • ymin (float) – Minimum y value of the map in world coordinates.

  • ymax (float) – Maximum y value of the map in world coordinates.

Side Effects:

Sets self.map_embedding: dense grid-map features. Sets self.ltn_embedding: compressed latent map features. Caches map boundary metadata such as self.xmin and self.xmax.

set_sur_info()[source]#

Extract local environment features at each pedestrian’s current position.

Maps pedestrian world coordinates in self.pos to raster-map indices and retrieves the corresponding vectors from the dense map embedding.

Side Effects:

Sets self.sur_info: environmental features under each pedestrian.

forward(denoise_t: torch.LongTensor, noisy_acc: torch.FloatTensor, ped_length: torch.LongTensor, veh_length: torch.LongTensor, timer: NamedTimer | None = None)[source]#

Forward pass: predict denoised trajectories from noisy inputs.

This method must be called after the set_*_embedding methods. It fuses the following information through Transformer decoder layers: 1. Diffusion timestep embedding (t) 2. Current noisy acceleration embedding (x_t) 3. Social interaction (Ped-Ped Attention) 4. Pedestrian-vehicle interaction (Ped-Veh Attention) 5. Environment interaction (Ped-Map Attention)

Parameters:

  • denoise_t (torch.LongTensor) – Current diffusion timestep t. Shape: (batch_size,)

  • noisy_acc (torch.FloatTensor) – Noisy future acceleration sequence, the diffusion input x_t. Shape: (batch_size, num_peds, pred_step, 2)

  • ped_length (torch.LongTensor) – Number of valid pedestrians per sample in the batch, used for masking. Shape: (batch_size,)

  • veh_length (torch.LongTensor) – Number of valid vehicles per sample in the batch, used for masking. Shape: (batch_size,)

  • timer (NamedTimer, optional) – Timer object for performance profiling.

Returns:

Model prediction.

If args.predict_noise is True, this is the predicted noise epsilon; otherwise it is the predicted original signal x_0 in acceleration space. Shape: (batch_size, num_peds, pred_step, 2)

Return type:

torch.FloatTensor

Contents