This paper presents a comparative analysis of algorithms for planning the pick sequence in a robotic sorting process involving objects moving on a conveyor belt. The primary objective was to determine the effects of decision-horizon length on computation time and solution quality and to evaluate the effects of conveyor speed and the number of sorted classes on capture rate (CR). A custom simulation environment was developed in Python using a simplified kinematic model of a delta robot. The analysis included exhaustive search, Branch-and-Bound, a time-bounded Monte Carlo method, and the FIFO, SPT, and NNF heuristics. Under the investigated conditions, factorial growth limited the computationally feasible exhaustive-search horizon to seven objects, whereas B&B extended exact planning to 13 objects with a computation time of 1.53 s. Monte Carlo was evaluated over 100 independent runs per horizon using a 1 s computation-time budget. All 700 runs achieved the same maximum number of seven successful picks as B&B, although the probability of reproducing the complete B&B optimum generally decreased for longer horizons. For the 20-object horizon, the mean and maximum increases in sequence execution time were 4.12% and 6.47%, respectively, while the planning time decreased from 27.29 s for B&B to 1 s for Monte Carlo. In the experiment comprising 5,000 objects, B&B achieved the highest CR in 38 of the 56 scenarios, whereas FIFO and SPT achieved the same or higher CR under lower process loads. These results provide a basis for selecting the planning algorithm according to the decision horizon and current process conditions.