Base station, roadside device, traffic communication system, traffic management method, and training data generation method
A base station used in a traffic communication system includes a communicator configured to receive, via wireless communication, information transmitted from a vehicle traveling on a road, and a controller configured to identify, based on the information received by the communicator, a traveling behavior including a travel route of the vehicle on the road. The controller is configured to execute first processing for performing statistical learning or machine learning on a plurality of the traveling behaviors respectively identified for a plurality of the vehicles traveling on the road, and second processing for determining, based on a result of the first processing, whether a traffic obstacle is present on the road and a position of the traffic obstacle on the road.
1. A traffic communication system comprising:
an image sensor configured to capture an image of a moving body on a road to output image data;
a receiver configured to receive, from the moving body via wireless communication, an uplink message including attribute data indicative of an attribute of the moving body;
a generator configured to associate the attribute data included in the uplink message with the image data output by the image sensor to generate training data to be used in construction of a learning model by machine learning, wherein
the learning model is a model that identifies an attribute of the moving body from the image data output by the image sensor; wherein
the image sensor is configured to capture an image of a predetermined geographic region on the road,
the uplink message includes position data indicating a geographic position of the moving body, and
the generator is configured to associate, in response to the position data included in the uplink message indicating the position of the predetermined geographic region, the attribute data included in the uplink message with the image data output by the image sensor to generate the training data.
2. A traffic communication system comprising:
an image sensor configured to capture an image of a moving body on a road to output image data;
a receiver configured to receive, from the moving body via wireless communication, an uplink message including attribute data indicative of an attribute of the moving body;
a generator configured to associate the attribute data included in the uplink message with the image data output by the image sensor to generate training data to be used in construction of a learning model by machine learning, wherein
the learning model is a model that identifies an attribute of the moving body from the image data output by the image sensor; wherein
the uplink message includes movement direction data indicating a movement direction of the moving body, and
the generator is configured to associate, in response to the movement direction data included in the uplink message indicating a predetermined movement direction, the attribute data included in the uplink message with the image data output by the image sensor to generate the training data.
3. A traffic communication system comprising:
an image sensor configured to capture an image of a moving body on a road to output image data;
a receiver configured to receive, from the moving body via wireless communication, an uplink message including attribute data indicative of an attribute of the moving body;
a generator configured to associate the attribute data included in the uplink message with the image data output by the image sensor to generate training data to be used in construction of a learning model by machine learning, wherein
the learning model is a model that identifies an attribute of the moving body from the image data output by the image sensor;
a light detection and ranging (LiDAR) sensor for detecting the moving body, wherein
the generator associates the attribute data included in the uplink message with the image data output by the image sensor and detection data output by the LiDAR sensor to generate the training data, and
the learning model is a model that identifies an attribute of the moving body from a combination of the image data output by the image sensor and the detection data output by the LiDAR sensor.