Algorithm to generate planning-based attention signals

A tangible, non-transitory, machine-readable medium, comprising machine-readable instructions, that when executed by a processor, cause the processor to:

• receive driving-scene data, wherein the driving-scene data is indicative of a driving scene around a host vehicle, and the driving-scene data includes map data and localization data, and the driving scene includes a plurality of actors;
• convert the driving-scene data into a scene-graph;
• input the scene-graph into a deep neural network (DNN);
• determine an attention score for each of the plurality of actors using the DNN, wherein the attention score of each of the plurality of actors represents a priority given to each of the plurality of actors based on the driving scene;
• and
• command the host vehicle to autonomously drive according to a trajectory determined taking into account the attention score of each of the plurality of actors, wherein the plurality of actors includes a removed actor and a plurality of remaining actors;
• creating a plurality of first scenarios by generating a plurality of first samples from an uncertainty distribution of a state of the plurality of actors;
• creating a plurality of second scenarios by generating a plurality of second samples from an uncertainty distribution of a state of the plurality of remaining actors;
• generating a first sample trajectory for each of the plurality of first scenarios;
• generating a second sample trajectory for each of the plurality of second scenarios;
• determining a first trajectory statistic using the first sample trajectory for each of the plurality of first scenarios;
• determining a second trajectory statistic using the second sample trajectory for each of the plurality of second scenarios;
• determining a distance from the first trajectory statistic to the second trajectory statistic;
• and
• wherein the first trajectory statistic is a first average trajectory, the second trajectory statistic is a second average trajectory, and the distance from the first trajectory statistic to the second trajectory statistic is determined using a following equation:
• D
• i
• =
• max
• t
• [
• (
• x
• i
• (
• t
• )
• -
• x
• o
• (
• t
• )
• )
• 2
• +
• (
• y
• i
• (
• t
• )
• -
• y
• 0
• (
• t
• )
• )
• 2
• ]
• where:
• t is time;
• (xo (t),yo (t)) are waypoints at time t of the first average trajectory;
• (xi(t),yi (t)) are waypoints at time t of second average trajectory;
• and
• Di is distance from the first average trajectory to the second average trajectory.