Object recognition device, object recognition method and self-driving system

Patent metadata

Field	Value
Publication number	US-2017372149-A1
Application number	US-201715623650-A
Country	US
Kind code	A1
Filing date	Jun 15, 2017
Priority date	Jun 24, 2016
Publication date	Dec 28, 2017
Grant date	—

Abstract

An object recognition device 4 includes: a data reception unit 5 that creates observation data of respective sensors 1, 2 in accordance with sensor's detection data of an object in the surroundings of a host vehicle; an association processing unit 6 which, based on an association reference value, generates association data denoting a correspondence between the observation data and object data of a previous process cycle; and an updating processing unit 7 which, based on the association data, updates a state vector included in the object data of the previous process cycle, and updates the object data by including latest association data being the observation data having corresponded to the object data most recently, wherein the association processing unit 6 generates the association reference value using preferentially the latest association data of the same sensor as that of the observation data of a current process cycle.

First claim

What is claimed is:

1 . An object recognition device that detects an object in the surroundings of a host vehicle using a plurality of sensors, and receives input of respective detection data constituting position information of the object from the sensors, the device comprising:
a data reception unit that creates and outputs observation data of each sensor, in accordance with the detection data from the plurality of sensors; an association processing unit which, on the basis of an association reference value, works out association data denoting a correspondence between the observation data input from the data reception unit and object data of a previous process cycle, and outputs the observation data and the association data; and an updating processing unit which, on the basis of the association data input from the association processing unit, updates a state vector included in the object data of the previous process cycle and pertaining to the position information of the object, updates, for each sensor, the object data by including, in the object data, latest association data that is the observation data having corresponded to the object data most recently, from among the observation data, and outputs the updated object data, wherein the association processing unit works out the association reference value using preferentially the latest association data of the same sensor as that of the observation data of a current process cycle.

2 . The object recognition device according to claim 1 , wherein the association processing unit goes back to a point in time at which the observation data corresponding to the object data in a past process cycle is received, and calculates a tentative state vector at the point in time, using a predicted value of the state vector of the current process cycle, and sets, as the association reference value, a value resulting from correcting the predicted value using a correction amount in the form of a difference between the tentative state vector and the observation data at the point in time.

3 . The object recognition device according to claim 1 , wherein when a detection count of the observation data of each sensor corresponding to the object data in a past process cycle is equal to or smaller than an integer established beforehand for each sensor, the association processing unit switches, in accordance with the detection count, the sensor that is used in order to work out the predicted value of the state vector.

4 . The object recognition device according to claim 1 , wherein at least one of the plurality of sensors is used as a sensor of a type different from the type of the other sensors.

5 . An object recognition method, comprising:
a first step of detecting an object in the surroundings of a host vehicle using a plurality of sensors, and receiving input of respective detection data constituting position information of the object from the sensors; a second step of creating and outputting observation data of each sensor, in accordance with the detection data from the plurality of sensors; a third step of, on the basis of an association reference value, working out association data denoting a correspondence between the observation data output in the second step and object data of a previous process cycle, and outputting the observation data and the association data; and a fourth step of, on the basis of the association data output in the third step, updating a state vector included in the object data of the previous process cycle and pertaining to the position information of the object, updating, for each sensor, the object data by including, in the object data, latest association data that is the observation data having corresponded to the object data most recently, from among the observation data, and outputting the updated object data, wherein, in the third step, the association reference value is worked out using preferentially the latest association data of the same sensor as that of the observation data of a current process cycle.

6 . A self-driving system that detects an object in the surroundings of a host vehicle using a plurality of sensors, and receives input of respective detection data constituting position information of the object from the sensors, the self-driving system comprising:
a data reception unit that creates and outputs observation data of each sensor, in accordance with the detection data from the plurality of sensors; an association processing unit which, on the basis of an association reference value, works out association data denoting a correspondence between the observation data input from the data reception unit and object data of a previous process cycle, and outputs the observation data and the association data; an updating processing unit which, on the basis of the association data input from the association processing unit, updates a state vector included in the object data of the previous process cycle and pertaining to the position information of the object, updates, for each sensor, the object data by including, in the object data, latest association data that is the observation data having corresponded to the object data most recently, from among the observation data, and outputs the updated object data; and a vehicle control unit that controls the operation of the host vehicle on the basis of the updated object data input from the updating processing unit, wherein the association processing unit works out the association reference value using preferentially the latest association data of the same sensor as that of the observation data of a current process cycle.

Assignees

• Mitsubishi Electric Corp

Inventors

• Mori Masanori

Classifications

• G06T7/277 Primary

  involving stochastic approaches, e.g. using Kalman filters · CPC title

• G06V20/58

  Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads · CPC title

• G06F18/25

  Fusion techniques · CPC title

• G06F18/256

  of results relating to different input data, e.g. multimodal recognition · CPC title

• G05D1/0088

  characterized by the autonomous decision making process, e.g. artificial intelligence, predefined behaviours (using knowledge based models G06N5/00) · CPC title