Multi-feature fusion series RNN structure and pedestrian prediction method

Abstract

The invention discloses a multi-feature fusion series RNN structure and a pedestrian prediction method. The structure comprises an information acquisition module, an information processing module, a series GRU module, a full connection layer module, an activation function module and a prediction module. The information acquisition module acquires video images of pedestrians and the surrounding environment when a vehicle is driven in environments with different roads and crowd densities and the speed of the vehicle. The information processing module processes the acquired data to generate a data set; each stage of GRU in the series GRU module processes different information in the data set and inputs the hidden state of the previous stage of GRU connected in series, and carries out fusion calculation on the different information; the full connection layer module integrates the multi-dimensional matrix to obtain a one-dimensional vector; the excitation function module processes the one-dimensional vector information; and the prediction module obtains a prediction result of the pedestrian trajectory. According to the method, information from multiple sources is fused layer by layer indifferent neural network layers according to the complexity of the information, and pedestrian behavior understanding and track prediction are realized.

Description

technical field [0001] The invention belongs to the field of computer vision for intelligent driving of automobiles, and specifically refers to a multi-feature fusion serial RNN (recurrent neural network) structure and a pedestrian prediction method. Background technique [0002] The rapid development of autonomous driving technology has put forward higher requirements for accurate understanding of pedestrian activities and prediction of pedestrian movement trajectories; accurate understanding of pedestrian activities and prediction of pedestrian movement trajectories will help the car driving system to choose the correct driving route and avoid potential human accidents. Vehicle collisions and the resulting interruption of traffic flow. In addition, neural networks have been widely used in the field of automatic driving for understanding pedestrian activities and predicting pedestrian trajectory. [0003] At present, the main method to solve the problem of pedestrian behav...

Images