Ahead of Chuseok, when parcel volumes surge, research and pilots are exploring AI, drones, and autonomous robots for delivery. Professor Minsoo Kim and his team in the School of Computing at KAIST developed an AI technology called “RL-SPH” that generates feasible delivery plans while accounting for constraints such as different vehicle cargo capacities and working hours. The results were presented at the International Conference on Machine Learning (ICML), held from July 6 to 11. RL-SPH first searches for a feasible plan satisfying all constraints, then improves it by reducing cost or time. Applied to five representative problems including delivery planning, it found feasible solutions for all, producing plans on average 28.6 times closer to the optimal solution, while the time needed to find the first feasible plan was cut to about one-fifth. The researchers expect the technology to apply to logistics and other domains requiring complex planning, such as manufacturing, semiconductor processing, and workforce allocation. Separately, the Korea AeroSpace Administration and the Electronics and Telecommunications Research Institute (ETRI) are developing a “drone–robot collaborative doorstep delivery system” linking drones capable of carrying up to 40 kg of cargo with delivery robots and docking stations.