发明名称 Method and system for predicting movement behavior of a target traffic object
摘要 A method for computationally predicting future movement behaviors of at least one target object can have the steps of producing sensor data by at least one sensor physically sensing the environment of a host vehicle, andcomputing a plurality of movement behavior alternatives of a target object sensed by the sensor(s).;The context based prediction step uses a set of classifiers, each classifier estimating a probability that said sensed target object will execute a movement behavior at a time. The method can also include validating the movement behavior alternatives by a physical prediction comparing measured points with trajectories of situation models and determining at least one trajectory indicating at least one possible behavior of the traffic participant,estimating at least one future position of the traffic participant based on the at least one trajectory, andoutputting a signal representing the estimate future position.
申请公布号 US8903588(B2) 申请公布日期 2014.12.02
申请号 US201213525496 申请日期 2012.06.18
申请人 Honda Research Institute Europe GmbH 发明人 Schmüdderich Jens;Rebhan Sven
分类号 G08G1/16;G06F7/00;G06F17/10;B60W30/095;G06K9/00;G06K9/62;B60W50/00;B60W30/16 主分类号 G08G1/16
代理机构 Squire Patton Boggs (US) LLP 代理人 Squire Patton Boggs (US) LLP
主权项 1. A method for computationally predicting future movement behaviors of at least one target object, said method comprising the steps of: producing sensor data by at least one sensor physically sensing the environment of a host vehicle, computing a plurality of movement behavior alternatives of a target object sensed by the sensor(s) for the same initial situation, by predicting movement behaviors of the target object applying a context based prediction step using indirect indicators derived from said sensor data, wherein said context based prediction step uses a set of classifiers for each movement behavior, each classifier estimating a probability that said sensed target object will execute a movement behavior at a time, wherein the probability is calculated based on a feature vector calculated for each classifier based on the indirect indicators, and wherein at least part of said indirect indicators relate to information about a relation between the target object and one or more other traffic participants, validating the movement behavior alternatives by a physical prediction comparing measured points with trajectories of situation models and determining at least one trajectory indicating at least one possible behavior of the target object, wherein the at least one movement behavior is a sub-set of the possible movement behavior alternatives, estimating at least one future position of the target object based on the at least one trajectory, and outputting a signal representing the estimated future position.
地址 Offenbach/Main DE