Robotics and Automation
Co-Principal Investigator
Dr. Ellips Masehian
Dr. Ellips Masehian is an associate professor at the Industrial and Manufacturing Engineering Department at Cal Poly Pomona. He received the B.S. and M.S. degrees in Industrial Engineering, both from Iran University of Science and Technology, Tehran, with honors, and a Ph.D. degree from Tarbiat Modares University. Dr. Masehian is one of the Co-PIs for the CREST-RASM grant.
Subproject 3
Motion, Grasp, and Regrasp Planning of Manipulator Robots for Agile Manufacturing and Assembly
Overview
Modern manufacturing demands robots capable of handling diverse parts and products with varying geometries, orientations, and movement trajectories. In fully automated smart factories, robots must grasp and regrasp both stationary and moving objects to load, unload, position, assemble, and transfer items efficiently. However, most current robots are preprogrammed for static objects and predefined paths, limiting their adaptability to dynamic, fast-paced environments. Despite significant research on robotic grasping, particularly for 5-fingered hands, much of it focuses on static systems, neglecting the critical need to grasp dynamic objects as they move on conveyor belts, roll on surfaces, or slide on slopes. Also, most works focus on just grasping the objects, often neglecting subsequent regrasping or reorienting operations. These limitations restrict robots from performing complex tasks and adapting to changing conditions, posing a significant barrier to the advancement of smart manufacturing.
Goal
The goal of this subproject is to address these gaps by developing methods and algorithms for autonomous robotic systems to perform advanced grasping and regrasping in dynamic settings.
By integrating techniques such as AI, sophisticated motion planning, and high-resolution sensors, the initiative aims to enhance robots’ real-time adaptability, precision, and efficiency. Leveraging innovative approaches, including haptic-based deep learning and massively parallel reinforcement learning, the project seeks to optimize robotic grasping and regrasping processes, meeting the demands of highly dynamic manufacturing environments.
The proposed Smart Factory integrates three layers:
- The Physical Layer, comprising manufacturing hardware like conveyors, CNC machines, a quality control station, ASRS, robotic arms, a mobile robot, and sensors
- The Digital Layer, a real-time simulation of the Physical Layer via 3D models connected to sensors
- The Virtual Layer, enabling user interaction with digital components through AR/VR technologies.
Outcome
The outcomes of this subproject will lead to a system capable of efficiently handling all material handling and assembly tasks while effectively responding to the dynamics and uncertainties inherent in a typical manufacturing environment.
Research Faculty and Staff
