Vehicle control device, vehicle control method, and storage medium
Provided is a vehicle control device including: a recognition unit configured to recognize an object which is located near a vehicle; a calculation unit configured to calculate an area of risk which is distributed around the object; a generation unit configured to generate a target trajectory for the vehicle to travel along on the basis of the area; and a driving control unit configured to automatically control at least one of a speed or steering of the vehicle on the basis of the target trajectory, wherein the generation unit selects one or a plurality of models from a plurality of models that output the target trajectory in a case where the area is input, inputs the area calculated by the calculation unit to the selected model, and generates the target trajectory on the basis of an output result of the model to which the area is input, and the calculation unit changes a range of the area in accordance with a type of model selected by the generation unit.
1. A vehicle control method comprising:
recognizing an object which is located near the vehicle;
calculating an area of risk which is distributed around the recognized object;
generating a target trajectory for the vehicle to travel along on the basis of the area of risk;
controlling at least one of a speed or steering of the vehicle on the basis of the generated target trajectory;
selecting one or a plurality of models from a plurality of models that output the target trajectory in a case where the area of risk is input, input the area of risk to the selected model, and generate the target trajectory on the basis of an output result of the model to which the area of risk is input; and
changing a size of a range of the area of risk to a different size in accordance with a type of the selected model among the plurality of models,
wherein the plurality of models include at least one first model which is rule-based or model-based and at least one second model which is machine-learning-based, and
wherein the range of the area of risk is set to a first range in a case where the model selected is the first model, the range of the area of risk is set to a second range smaller than the first range in a case where the model selected is the second model, and calculating a potential of the risk within the set range.
2. A vehicle control device comprising:
a processor configured to:
recognize an object which is located near a vehicle;
calculate an area of risk which is distributed around the object;
generate a target trajectory for the vehicle to travel along on the basis of the area of risk;
automatically control at least one of a speed or steering of the vehicle on the basis of the target trajectory;
select one or a plurality of models from a plurality of models that output the target trajectory in a case where the area of risk is input, input the area of risk to the selected model, and generate the target trajectory on the basis of an output result of the model to which the area of risk is input; and
change a size of a range of the area of risk to a different size in accordance with a type of model among the plurality of models,
wherein the plurality of models include at least one first model which is rule-based or model-based and at least one second model which is machine-learning-based, and
wherein, the processor is further configured to set the range of the area of risk to a first range in a case where the model selected is the first model, set the range of the area of risk to a second range smaller than the first range in a case where the model selected is the second model, and calculate a potential of the risk within the set range.
3. The vehicle control device according to claim 2 , wherein the processor is further configured to select any one model from the plurality of models in accordance with a type of road on which the vehicle travels.
4. The vehicle control device according to claim 3 , wherein the processor is further configured to select the first model in a case where the vehicle travels on an expressway, and select the second model in a case where the vehicle travels on a road with more obstacles than an expressway, has a complicated surrounding situation, or a road with unrecognizable lanes.
5. The vehicle control device according to claim 2 , wherein the processor is further configured to select any one model from the plurality of models in accordance with a level of automation when at least one of the speed or steering of the vehicle is controlled automatically.
6. The vehicle control device according to claim 5 , wherein the processor is further configured to select the first model in a case where the level is equal to or higher than a reference value, and select the second model in a case where the level is lower than the reference value.
7. The vehicle control device according to claim 2 , wherein the processor is further configured to input the area of risk to each of the plurality of models, generate a plurality of target trajectories on the basis of an output result of each of the plurality of models to which the area of risk is input, and the target trajectory closest to a trajectory when a person drives the vehicle or another vehicle from the plurality of generated target trajectories, and
control at least one of the speed or steering of the vehicle on the basis of the target trajectory.
8. The vehicle control device according to claim 2 , wherein the area of risk is an area partitioned by a plurality of meshes,
the potential of the risk calculated on the basis of a state of at least one of the vehicle or the object is associated with each of the plurality of meshes,
the potential of the risk associated with each of the plurality of meshes is normalized on the basis of potentials of all meshes included in the area of risk to generate a normalized potential, and
the processor is further configured to generate a trajectory passing through the mesh of which the normalized potential is lower than a threshold as the target trajectory.
9. A computer readable non-transitory storage medium having a program stored therein, the program causing a computer mounted in a vehicle to execute:
recognizing an object which is located near the vehicle;
calculating an area of risk which is distributed around the recognized object;
generating a target trajectory for the vehicle to travel along on the basis of the area of risk;
controlling at least one of a speed or steering of the vehicle on the basis of the generated target trajectory;
selecting one or a plurality of models from a plurality of models that output the target trajectory in a case where the area of risk is input, inputting the area of risk to the selected model, and generating the target trajectory on the basis of an output result of the model to which the area of risk is input; and
changing a size of a range of the area of risk to a different size in accordance with a type of the selected model among the plurality of models,
wherein the plurality of models include at least one first model which is rule-based or model-based and at least one second model which is machine-learning-based, and
wherein the range of the area of risk is set to a first range in a case where the model selected is the first model, the range of the area of risk is set to a second range smaller than the first range in a case where the model selected is the second model, and calculating a potential of the risk within the set range.