This project aims to develop a robotic pick-and-place system that goes beyond traditional methods by intelligently planning and executing object placement in specific poses, addressing complex constraints found in industrial tasks like kitting and assembly. At its core are automated planning and perception: the planning module generates plans based on predefined constraints, while the perception module detects and estimates the poses of objects and containers. A key task example involves placing objects by specified shape and color into corresponding containers. The system is divided into detection, planning, and in-hand perception and control modules. These work in tandem to identify object properties, generate step-by-step tasks from natural language inputs, and adjust the robot arm's movement to ensure objects are correctly oriented to match container specifications. Each module has been evaluated individually and has been integrated into a complete proof-of-concept system. The system demonstrates the feasibility of visual reasoning for pick and place, including in-hand perception to account for uncertainty in the attained grasp configuration.