The application relates to the technical field of path planning, and discloses a path optimization method for multi-vehicle cyclic goods taking three-dimensional packing, a terminal and a medium. The method acquires relevant data, maps the part requirement of a supplier into a
box set in units of packaging boxes, and determines a unit
residual space opportunity cost coefficient; an original model is constructed according to the relevant data, and the objective function of the original model contains
vehicle driving cost, vehicle starting fixed cost and
residual space opportunity cost of a
carriage; the original model is decomposed into a limited main problem and a pricing sub-problem by using a Dantzig-Wolfe
decomposition method; the limited main problem and the pricing sub-problem are iteratively solved by using a
column generation algorithm; if the solution of the limited main problem contains a non-integer, a
branch and bound process is entered, the
column generation algorithm is repeatedly executed at a
branch node, and the process is terminated until a preset termination condition is met, and an optimal cyclic goods taking scheme including vehicle path planning and
carriage three-dimensional packing
layout is output. The application can improve the transportation efficiency of cyclic goods taking.