The invention relates to the technical field of intelligent logistics and
mobile robot autonomous navigation, in particular to an industrial park unmanned vehicle multi-task cooperative
processing system based on visual intelligence, and the
system comprises a global
potential energy field construction center which calls task distribution information of a to-be-processed task, and builds global
gravitational field data containing the
gravitational potential energy distribution of a task target point; the local semantic
perception unit is used for collecting real-time environment image data in the driving process of the unmanned vehicle to obtain environment
semantic feature data containing obstacle categories and
road surface texture attributes; the field parameter
coupling unit is used for generating local correction
field data; the
potential energy synthesis unit is used for obtaining synthesized
potential energy surface data; the gradient driving
control unit is used for calculating a negative gradient vector of the synthesized
potential energy surface data at the current position of the unmanned vehicle and converting the negative gradient vector into an
acceleration control instruction to drive the unmanned vehicle to move; according to the method, the communication overhead and the
deadlock probability are remarkably reduced, and the response speed and the environmental adaptability of the large-scale cluster operation are improved.