def __init__(self,
embedding_dim=64,
h_dim=64,
mlp_dim=1024,
bottleneck_dim=1024,
activation='relu',
batch_norm=True,
dropout=0.0,
pooling_dim=2,
neighborhood_size=2.0,
pool_every=False):
super(PoolHiddenNet, self).__init__()
self.mlp_dim = mlp_dim
self.h_dim = h_dim
self.bottleneck_dim = bottleneck_dim
self.embedding_dim = embedding_dim
self.pooling_dim = pooling_dim
self.neighborhood_size = neighborhood_size
mlp_pre_dim = embedding_dim + h_dim
mlp_pre_pool_dims = [mlp_pre_dim, self.mlp_dim * 8, bottleneck_dim]
self.spatial_embedding = nn.Linear(pooling_dim,
embedding_dim).to(device)
self.mlp_pre_pool = make_mlp(mlp_pre_pool_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout)
def __init__(self,
h_dim=64,
bottleneck_dim=1024,
activation='relu',
batch_norm=True,
dropout=0.0,
neighborhood_size=2.0,
grid_size=8,
pool_dim=None):
super(SocialPoolingAttention, self).__init__()
self.h_dim = h_dim
self.grid_size = grid_size
self.neighborhood_size = neighborhood_size
self.attention_decoder = Attention_Decoder(
attention_dim=bottleneck_dim,
embed_dim=4,
decoder_dim=h_dim,
encoder_dim=h_dim)
mlp_pool_dims = [bottleneck_dim, bottleneck_dim]
self.mlp_pool = make_mlp(mlp_pool_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout).to(device)
def __init__(self,
obs_len,
pred_len,
embedding_dim=64,
h_dim=64,
mlp_dim=1024,
num_layers=1,
activation='relu',
batch_norm=True,
dropout=0.0,
d_type='local',
grid_size=8,
neighborhood_size=2.0):
super(TrajectoryDiscriminator, self).__init__()
self.obs_len = obs_len
self.pred_len = pred_len
self.seq_len = obs_len + pred_len
self.mlp_dim = mlp_dim
self.h_dim = h_dim
self.d_type = d_type
self.encoder = Encoder(embedding_dim=embedding_dim,
h_dim=h_dim,
mlp_dim=mlp_dim,
num_layers=num_layers,
dropout=dropout)
real_classifier_dims = [h_dim, mlp_dim, 1]
self.real_classifier = make_mlp(real_classifier_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout)
def __init__(self,
obs_len,
pred_len,
embedding_dim=64,
encoder_h_dim=64,
decoder_h_dim=128,
mlp_dim=1024,
num_layers=1,
noise_dim=(0, ),
noise_type='gaussian',
noise_mix_type='ped',
dropout=0.0,
activation='relu',
batch_norm=True,
pooling=None,
pooling_output_dim=64,
decoder=None):
super(TrajectoryGenerator, self).__init__()
self.obs_len = obs_len
self.pred_len = pred_len
self.mlp_dim = mlp_dim
self.encoder_h_dim = encoder_h_dim
self.decoder_h_dim = decoder_h_dim
self.embedding_dim = embedding_dim
self.noise_dim = noise_dim
self.num_layers = num_layers
self.noise_type = noise_type
self.noise_mix_type = noise_mix_type
self.noise_first_dim = 0
# pooling options
self.pooling = pooling
self.pooling_output_dim = pooling_output_dim
self.encoder = Encoder(embedding_dim=embedding_dim,
h_dim=encoder_h_dim,
mlp_dim=mlp_dim,
num_layers=num_layers,
dropout=dropout)
self.decoder = decoder
if self.noise_dim[0] == 0:
self.noise_dim = None
else:
self.noise_first_dim = noise_dim[0]
mlp_input_dim = self.pooling_output_dim
if self.mlp_decoder_needed():
mlp_decoder_context_dims = [
mlp_input_dim, mlp_dim, decoder_h_dim - self.noise_first_dim
]
self.mlp_decoder_context = make_mlp(mlp_decoder_context_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout)
def __init__(
self, obs_len, pred_len, embedding_dim=64, h_dim=64, mlp_dim=1024,
num_layers=1, activation='relu', batch_norm=True, dropout=0.0,
c_type='local', collision_threshold=.25, occupancy_threshold=1.0,
down_samples=200, pooling=None, pooling_output_dim=64
):
super(TrajectoryCritic, self).__init__()
self.obs_len = obs_len
self.pred_len = pred_len
self.seq_len = obs_len + pred_len
self.mlp_dim = mlp_dim
self.h_dim = h_dim
self.activation = activation
self.c_type = c_type
self.collision_threshold = collision_threshold
self.occupancy_threshold = occupancy_threshold
# pooling options
self.noise_first_dim = 0
self.pooling = pooling
self.pooling_output_dim=pooling_output_dim
self.encoder = Encoder(
embedding_dim=embedding_dim,
h_dim=h_dim,
mlp_dim=mlp_dim,
num_layers=num_layers,
dropout=dropout
)
real_classifier_dims = [self.pooling_output_dim, mlp_dim, 1]
self.real_classifier = make_mlp(
real_classifier_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout)
mlp_dims = [self.pooling_output_dim, mlp_dim, 64]
self.mlp = make_mlp(
mlp_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout)
def __init__(self,
seq_len,
embedding_dim=64,
h_dim=128,
mlp_dim=1024,
num_layers=1,
pool_every_timestep=True,
dropout=0.0,
activation='relu',
batch_norm=True,
pool_static_type='None',
pooling=None,
pooling_output_dim=64):
super(Decoder, self).__init__()
self.seq_len = seq_len
self.mlp_dim = mlp_dim
self.h_dim = h_dim
self.embedding_dim = embedding_dim
# pooling options
if pool_static_type and pool_static_type.lower() == 'none':
pool_static_type = None
self.pool_static_type = pool_static_type
self.pool_every_timestep = pool_every_timestep
self.pooling = pooling
self.pooling_output_dim = pooling_output_dim
self.decoder = nn.LSTM(embedding_dim,
h_dim,
num_layers,
dropout=dropout)
if pool_every_timestep:
mlp_dims = [self.pooling_output_dim, mlp_dim, h_dim]
self.mlp = make_mlp(mlp_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout)
self.spatial_embedding = nn.Linear(2, embedding_dim).to(device)
self.hidden2pos = nn.Linear(h_dim, 2)
def __init__(self,
h_dim=64,
bottleneck_dim=1024,
activation='relu',
batch_norm=True,
dropout=0.0,
neighborhood_size=2.0,
grid_size=8,
pool_dim=None):
super(SocialPooling, self).__init__()
self.h_dim = h_dim
self.grid_size = grid_size
self.neighborhood_size = neighborhood_size
if pool_dim:
mlp_pool_dims = [grid_size * grid_size * h_dim, pool_dim]
else:
mlp_pool_dims = [grid_size * grid_size * h_dim, bottleneck_dim]
self.mlp_pool = make_mlp(mlp_pool_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout)
def __init__(self,
embedding_dim=64,
h_dim=64,
mlp_dim=1024,
bottleneck_dim=1024,
activation='relu',
batch_norm=True,
dropout=0.0,
grid_size=8,
neighborhood_size=2.0,
pool_static_type='random',
down_samples=200):
super(GridPooling, self).__init__()
self.mlp_dim = mlp_dim
self.h_dim = h_dim
self.bottleneck_dim = bottleneck_dim
self.embedding_dim = embedding_dim
self.neighborhood_size = neighborhood_size
self.pool_static_type = pool_static_type
self.grid_size = grid_size
self.encoder_dim = 1
self.static_scene_feature_extractor = StaticFeatures(down_samples)
self.attention_decoder = Attention_Decoder(
attention_dim=bottleneck_dim,
embed_dim=4,
decoder_dim=h_dim,
encoder_dim=1)
mlp_pool_dims = [bottleneck_dim, bottleneck_dim]
self.mlp_pool = make_mlp(mlp_pool_dims,
activation=activation,
batch_norm=batch_norm,
dropout=dropout).to(device)