speed: [B*N, T] pos_x: [B*N, T] pos_y: [B*N, T] already_stopped: [B*N, T] rearrange the signals to match the

arrange the signals according to the ordering prescribed by the stl formula

Update tracked agents object such that: - if agent was not there -> add it - if agent is not here anymore -> remove it This wi

Vectorize agents in a frame. Arguments: selected_track_id: selected_track_id: Either None for AV, or the ID of an agent that you

A factory for creating training algos Args: config (ExperimentConfig): an ExperimentConfig object, modality_shapes (dict): a

:param theta1: angle 1 (..., 1) :param theta2: angle 2 (..., 1) :return diff: smallest angle difference between angles (..., 1)

Append list of episodic experience Args: episodes_data (list): a list of episodic experiences

Ensure that array or tensor @x has size @size in dim @dim. Args: x (np.ndarray or torch.Tensor): input array or tensor size

Compute 'Scaled Dot Product Attention

Try to translate registered name to maintain backward compatibility.

Compute a negative log-likelihood for the multi-modal scenario. log-sum-exp trick is used here to avoid underflow and overflow, For more info

Return the probability of the true mode Args: ground_truth (np.ndarray): array of shape (timesteps)x(2D coords) pred (np.ndar

Simple differentiable bicycle model that does not allow reverse Args: state (torch.Tensor): a batch of current kinematic state [B, ..

Return bilinear interpolation of 4 nearest pts w.r.t to x,y from img Args: img (torch.Tensor): Tensor of size cxwxh. Usually one channel o

Return bilinear interpolation of 4 nearest pts w.r.t to x,y from img Args: img (torch.Tensor): Tensor of size cxwxh. Usually one channel o

Choose actions based on the guide losses -act_idx: (B, )

Compute histogram statistics for a run

Creates guidance configs for each scene based on the given configuration.

Compute KL Divergence loss Args: outputs (dict): outputs of the self.forward() call Returns: a dict

Compute KL Divergence loss Args: outputs (dict): outputs of the self.forward() call Returns: a dict

Compute per-agent and per-scene average speed

Compute per-agent collision rate and type

Compute per-agent collision rate and type

stores all given positions for needed agents

stores all given positions for needed agents

stores all given positions for needed agents

stores all given positions for needed agents

stores all given positions for needed agents

Compute per-agent and per-scene average speed

Compute per-agent and per-scene average speed

Compute per-agent and per-scene average speed violation

Compute per-agent and per-scene average acc violation

Compute per-agent and per-scene average metric

Compute per-agent and per-scene collision rate and type

Compute per-agent and per-scene collision rate and type