In common industrial settings, manipulators operate in structured environments with minimal uncertainties. Modern production demands require “smart industrial robots” capable of handling objects in unstructured and dynamic settings. While existing grasp planning methods work well in controlled environments, they struggle with uncertain scenarios, occlusions, and dynamic changes. These limitations affect real-world tasks like bin picking, decluttering, and assembly. This seminar will present planning approaches that leverage optimization and learning-based techniques to enable effective robotic manipulation in unstructured environments, with a focus on bimanual handover planning and push-grasp coordination.