The invention discloses a multi-parking-lot multi-vehicle-model vehicle path planning method. The method comprises the following steps: 1) establishing an objective function by taking the lowest totalcost of vehicle distribution as an objective; 2) encoding and redefining an operation mode; 3) initializing a population; 4) calculating the fitness of the artificial fish, and recording the optimalindividual; 5) evaluating each fish, selecting the behavior of each fish at the same time, and selecting the behavior with the highest fitness improvement to execute; 6) calculating the fitness of theobtained new population, and updating the optimal solution in the population; 7) performing annealing search on the optimal solution of the population; 8) ending the primary iteration; 9) judging whether a termination condition is met or not, and if so, executing the step 10; otherwise, returning to execute the step 5, and carrying out a new round of iteration; and 10) decoding the finally recorded optimal individual, calculating the reciprocal of the fitness of the optimal individual, i.e., the value of the target function, and after decoding, taking the client between two same vehicle numbers in the obtained decoding sequence as the distribution route of the vehicle, and outputting an optimal scheme when the unused vehicle does not appear in the decoding sequence.