Abstract: The selection of operation node sequence has a significant influence on the operation distance and operation time of harvesters, especially those with capacity constraints such as limited grain tank size. To reduce the operation distance and operation time during rice harvesting, in this paper, a multi-objective optimization model is presented with the variable of operation node sequence. The operation distance consists of rice harvesting distance and non-working distance. The operation time includes the time for rice harvesting and the time spent on non-working activities, such as returning to the unloading point. The constraints such as capacity of grain tank, avoiding repeating path are included in the multi-objective optimization model to ensure the effectiveness and validation of proposed model. During the rice harvesting process, each row of rice must be harvested completely at once. When the harvester reaches the maximum capacity of the grain tank or the remaining capacity of the grain tank cannot accommodate the next row of rice, the harvester must return to the unloading point. There is only one unloading point in each field and its location is the same as the position where the harvester enters the field. To solve this multi-objective optimization model, an improved ant colony optimization (IACO) algorithm is developed based on the traditional ant colony optimization (ACO) algorithm. In the IACO algorithm, each ant represents a potential solution. Through iterative exploration and pheromone updating, the algorithm converges toward an optimal or near-optimal solution. To avoid stagnation during the search process, the IACO integrates a selection principle combining deterministic and stochastic methods. The deterministic principle selects the option with the highest probability, while the stochastic principle employs the roulette wheel method to introduce randomness. This hybrid approach allows the algorithm to escape local optima, reduce runtime, and accelerate convergence. Additionally, the pheromone update strategy is refined to update only the pheromones on the best-performing ant's path. According to the TOPSIS (technique for order preference by similarity to an ideal solution) evaluation method, the optimal weighting scheme is that the weight of operation distance is 0.6 and the weight of operation time is 0.4. The parameters of the IACO algorithm are determined through a sensitivity analysis of each factor. The number of ants is 120, the pheromone factor is 1, the heuristic function factor is 3, and the pheromone volatilization factor is 0.3. Simulation experiments using the traditional ACO and the IACO are conducted to validate the model. From the optimization results it can be found that IACO reduces operation distance and standard deviation of operation distance by 5.47% and 71%, respectively. Moreover, the IACO also can decrease operation time by 5.29% and standard deviation of operation time by 47%, respectively. The algorithm running time and its standard deviation are decreased by 12.25% and 37.07%, respectively. These results highlight the superiority and stability of IACO. Field experiments further demonstrate that the proposed multi-objective optimization model and algorithm can reduce the operation distance by 9.5% and operation time by 9.1% in trapezoidal field when compared to the traditional ox-plowing method. In rectangular field, the operation distance can be reduced by 2.6% and the operation time can be saved by 6.9%. The proposed multi-objective optimization model and the IACO algorithm are effective for path planning of harvesters, which provide a good solution for intelligent multi-machine collaborative harvesting operation.