Autonomous mobile platform environment perception and mapping method based on bionics

Abstract

The invention discloses an autonomous mobile platform environment perception and mapping method based on bionics, and the method comprises the following steps: S1, collecting an automatic driving dataset, transmitting a picture to an improved semantic segmentation network, and carrying out the training of the improved semantic segmentation network; S2, initializing the combination of binocular vision and IMU; S3, fusing the binocular vision and the IMU; S4, constructing a bionic cell model; S5, constructing a cognitive map. An empirical map is established by using a bionics principle, so thatthe parameter quantity in the map is greatly reduced, and scene map information with a relatively large range can be stored. And meanwhile, modeling site information is semantically segmented and fused with position cells, so that the detection robustness can be improved. The strong feature extraction and learning capacity of the convolutional neural network is fully utilized, the problems that atraditional SLAM method is too large in mapping calculation amount and inaccurate in recognition can be solved, and the accuracy of location judgment in the bionic environment perception and mappingprocess is well improved.

Description

technical field [0001] The invention belongs to the technical field of positioning and navigation of an autonomous mobile platform, and in particular relates to a bionics-based environment perception and mapping method for an autonomous mobile platform. Background technique [0002] Simultaneous Localization and Mapping (SLAM) refers to a technology in which a robot actively creates a map in an unknown environment and performs self-positioning based on the estimation of its own state and the map. Accurate positioning information helps autonomous mobile platforms to complete tasks such as path planning and map drawing. Traditional real-time positioning and map construction algorithms use a method based on Bayesian probability estimation to calculate velocity and pose by collecting external input information. And use complex nonlinear optimization knowledge to solve the prior information in the environment. Build a huge database as a search source. However, when building a ...

AI Technical Summary

Problems solved by technology

However, when building a model in a real environment with complex environmental characteristics, it needs to consume a huge amount of computing power.

And the resulting modeling results cannot be used for deeper navigation tasks such as semantic analysis and cognitive understanding

Images