The application provides a 4WID-4WIS overdrive
mobile robot motion control system, which comprises an environment
perception module, a behavior decision module and a
motion control module; the environment
perception module is used for acquiring current state information and surrounding environment information of the
robot and constructing a unified state description based on a multi-
sensor fusion mode; the behavior decision module takes the
state vector output by the environment
perception module as input, outputs a current optimal action
mode number through a strategy network, and corresponds to a predefined
motion control mode; the motion control module is internally provided with multiple motion control
modes, each mode corresponding to an
independent motion controller; after receiving the action
mode number output by the behavior decision module, the motion control module is automatically switched to the motion
control mode and outputs
executable bottom-layer control instruction signals. The application adopts a
reinforcement learning mode to make behavior decisions, avoids precise modeling of a complex overdrive
system, and reduces the influence of modeling errors on control performance.