Ανάπτυξη συστήματος εκτίμησης βέλτιστων σημείων σύλληψης σε εφαρμογές ρομποτικού χειρισμού αντικειμένων

In bin-picking applications that rely on suction grasping, it’s important to ensure optimized picking approach that securely lifts the part and allows for dynamic manipulation. Most grasp planners address geometric feasibility and static stability, but do not consider robustness under dynamic reorientation or consider multi-cup configurations. This work presents an integrated tool that automatically generates, evaluates, and visualizes suction grasp candidates directly from CAD models of parts in cluttered environments. The proposed framework accepts physical parameters of the suction cups, the target object, vacuum pressure, suction configuration and reorientation and a 3D mesh representation of the part. It computes feasible suction contact locations and multi-cup configurations, evaluates their ability to resist external wrenches under gravity and rotational motion, and estimates suction cup deformation during object reorientation to produce an estimation of grasp stability and a visualization of target object final pose. This framework is evaluated under an elevators manufacturing case, where a physical robotic system utilizes the generated grasp candidates and expected part behavior towards the accurate bin picking for electrical panel assembly.