Method for adaptive identification of erroneous GPS observed value

A method for adaptive identification of an erroneous GPS observed value, comprising:

• acquiring consecutive positioning information of a vehicle from a GPS sensor of the vehicle;
• extracting first observed value data from the positioning information, wherein the first observed value data comprises respective data on data status value, heading significant bit, the number of satellites used and horizontal dilution of precision;
• acquiring posture information and speed information of the vehicle;
• deriving dead reckoning trajectory data of the vehicle from calculation based on the posture information and the speed information;
• marking a level of the first observed value data as a first level, when the data status value of the first observed value data is equal to a preset expected data status value, the heading significant bit is valid, the number of satellites used is greater than a preset threshold of number of satellites used, and the horizontal dilution of precision is less than a preset threshold of horizontal dilution of precision;
• marking the level of the first observed value data as a second level, when the data status value of the first observed value data is equal to the expected data status value, the heading significant bit is invalid, the number of satellites used is greater than the threshold of number of satellites used, and the horizontal dilution of precision is less than the threshold of horizontal dilution of precision;
• acquiring a set of levels of the first observed value data for a preset number of consecutive frames prior to current first observed value data;
• marking the current first observed value data as second observed value data, when each level in the set of levels is the first level;
• marking the current first observed value data as the second observed value data, when the set of levels only include the first level and the second level and the amount of the first observed value data at the second level is less than a preset threshold of amount for the second level;
• setting a covariance matrix and a kernel function, by taking a matching position of a position sensor of the vehicle as a vertex to which one piece of the second observed value data is added and taking a first relative observable of the position sensor at first consecutive time instants and a second relative observable of the dead reckoning trajectory data at second consecutive time instants as constraint edges, to thereby construct pose graph data;
• deriving processing result information from calculation based on the pose graph data;
• determining for each second observed value data whether a cost function of the second observed value data exceeds a preset cost function threshold, based on the processing result information;
• marking the second observed value data, of which the cost function is not greater than the cost function threshold, as third observed value data, when the amount of the second observed value data, of which the cost function exceeds the cost function threshold, is less than a preset second amount threshold;
• and
• acquiring real-time positioning data by processing the third observed value data and three-dimensional scene map data.