Object recognition integration device and object recognition integration method
View Patent ↗Provided are an object recognition integration device and an object recognition integration method, which are capable of integrating pieces of detection data that are detected by a respective plurality of sensors in consideration of an inaccuracy of identification of objects. An association relationship between measurement data and previous object data is determined based on an object type and a certainty for each object type candidate contained in measurement data generated for each of the plurality of sensors, and an object type and a certainty for each object type candidate contained in the previous object data. Then, association data is generated by associating the measurement data and the previous object data, which are determined as having “possibility of association”, with each other, to thereby generate current object data by updating the previous object data with use of the association data.
1. An object recognition integration device, comprising:
at least one processor which implements:
a data reception unit configured to:
receive pieces of detection data from a respective plurality of sensors;
generate measurement data for each of the respective plurality of sensors based on the received pieces of detection data; and
output the measurement data for each of the respective plurality of sensors;
an association data generation unit configured to:
determine an association relationship between the measurement data output from the data reception unit and previous object data, which is object data generated at a processing time previous to a current processing time based on previous measurement data output from the data reception unit at the processing time previous to the current processing time, as any one of a possibility of association and no possibility of association based on an object type and a certainty for each object type candidate contained in the measurement data and an object type and a certainty for each object type candidate contained in the previous object data;
associate the measurement data and the previous object data, which are determined as having the possibility of association, with each other;
generate association data in which the measurement data, the previous object data, and the association therebetween are stored; and
output the generated association data; and
an object data generation unit configured to:
generate current object data, which is object data at the current processing time, by updating a state of an object, the object type, and the certainty contained in the previous object data based on the association data output from the association data generation unit; and
output the generated current object data to a vehicle control unit configured to perform automatic driving control of a vehicle based on the current object data,
wherein the association data generation unit is further configured to:
perform association possibility determination with use of a residual vector for the association relationship between the measurement data and the previous object data; and
determine the association relationship by correcting the association possibility determination based on the object type and the certainty contained in the measurement data and the object type and the certainty contained in the previous object data.
2. An object recognition integration device, comprising:
at least one processor which implements:
a data reception unit configured to:
receive pieces of detection data from a respective plurality of sensors;
generate measurement data for each of the respective plurality of sensors based on the received pieces of detection data; and
output the measurement data for each of the respective plurality of sensors;
an association data generation unit configured to:
determine an association relationship between the measurement data output from the data reception unit and previous object data, which is object data generated at a processing time previous to a current processing time based on previous measurement data output from the data reception unit at the processing time previous to the current processing time, as any one of a possibility of association and no possibility of association based on an object type and a certainty for each object type candidate contained in the measurement data and an object type and a certainty for each object type candidate contained in the previous object data;
associate the measurement data and the previous object data, which are determined as having the possibility of association, with each other;
generate association data in which the measurement data, the previous object data, and the association therebetween are stored; and
output the generated association data; and
an object data generation unit configured to:
generate current object data, which is object data at the current processing time, by updating a state of an object, the object type, and the certainty contained in the previous object data based on the association data output from the association data generation unit; and
output the generated current object data to a vehicle control unit configured to perform automatic driving control of a vehicle based on the current object data,
wherein the association data generation unit is further configured to associate the previous object data with measurement data having a smallest weighted average of residual vectors between the previous object data and the measurement data among pieces of the measurement data determined as having the possibility of association with the previous object data.
3. An object recognition integration method, comprising:
a data reception step of:
receiving pieces of detection data from a respective plurality of sensors;
generating measurement data for each of the respective plurality of sensors based on the received pieces of detection data; and
outputting the measurement data for each of the respective plurality of sensors;
an association data generation step of:
determining an association relationship between the measurement data output in the data reception step and previous object data, which is object data generated at a processing time previous to a current processing time based on previous measurement data output in the data reception step at the processing time previous to the current processing time, as any one of a possibility of association and no possibility of association based on an object type and a certainty for each object type candidate contained in the measurement data and an object type and a certainty for each object type candidate contained in the previous object data;
associating the measurement data and the previous object data, which are determined as having the possibility of association, with each other;
generating association data in which the measurement data, the previous object data, and the association therebetween are stored; and
outputting the generated association data; and
an object data generation step of:
generating current object data, which is object data at the current processing time, by updating a state of an object, the object type, and the certainty contained in the previous object data based on the association data output in the association data generation step; and
outputting the generated current object data to a vehicle control unit that performs automatic driving control of a vehicle based on the current object data,
wherein the association data generation step further comprises associating the previous object data with measurement data having a smallest weighted average of a residual vector between the previous object data and the measurement data and a residual of the certainty between the previous object data and the measurement data among pieces of the measurement data determined as having the possibility of association with the previous object data.