Flexible Manufacturing Line with Multiple Robotic Cells Heping Chen hc15@txstate.edu Robotics Laboratory Ingram School of Engineering Texas State University 1 Robotics Laboratory Ingram School of Engineering Bring Ideas Together OUTLINE • Introduction • Autonomous Robot Teaching – POMDP based learning algorithm • Performance Optimization – Single cell optimization – Multi cell Optimization • Discussion 2 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Trim and Final Assembly Line 3 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Problems in Production Line • When one work station fails, the production line has to be stopped – Low efficiency – High cost • For complex manufacturing processes, is it possible to optimize the system performance? – Single work station optimization? – Multi work station optimization? 4 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Proposed Solution Station controller/ Computer Station controller/ Computer Station controller/ Computer Other devices Station controller/ Computer Station controller/ Computer 5 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Problems • Big part location errors – Can happen after a new batch is launched – Parts could come from different suppliers • Typically manual teaching methods are used to adjust the parameters – Production line has to be stopped • Installing additional sensors generates new problems – System upgrading cost – Maintenance issues – Sensor failure 6 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Failure Correction Online Transfer Learning Online Transfer Learning Online Learning Online Learning #1 FMS #2 Online Learning FMS #n f-L Of L noO et in eT in ea g in ch f-L f O MTS FMS g in h c ea T e in L nO o et n i Kinect SICK FMS:Fixed Manufacturing Station MTS:Mobile Teaching Station 7 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Autonomous Robot Teaching • • • • Reducing the operational cost Sensor accuracy problem Calibration is difficult Noise Target Hole Marker for Tool Detection Transmission Part Robot Tool 8 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Autonomous Robot Teaching Autonomous Industrial Mobile Manipulator ABB IRB 140 Industrial Camera Hole Transmission Part “Adult” Robot PC Controller Tool/Peg Gigabit Ethernet Ethernet DALSA Industrial Camera IRC5 Robot Controller 9 Workpiece Tool & Peg “Child”IRB140 Robotics Laboratory Ingram School of Engineering Bring Ideas Together POMDP • Partial Observable Markov Decision Process – State Transition is Uncertain – Observation is uncertain – State is partial observable • Why POMDP? – Using POMDP to estimate the underlying errors through executing actions and receiving observations 10 Robotics Laboratory Ingram School of Engineering Bring Ideas Together POMDP Method • Belief State – – – – Real state is unknown How to make decision? Assign a belief b over state S Update the belief state in each step – Defining Value Functions – Approximating Value Functions • Alpha Vector 11 Robotics Laboratory Ingram School of Engineering Bring Ideas Together OUTLINE • Introduction • Autonomous Robot Teaching – MDP based learning algorithm – POMDP based learning algorithm • Performance Optimization – Single work station optimization – Multi work station optimization • Discussion 12 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Example: Single Stage Assembly 13 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Assembly Process Parameters: search force, search radius, search speed, insertion force How to tune the Assembly Parameters? 14 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Example: Multi-Stage Assembly 15 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Problems • Design of Experiment (DOE) • Genetic Algorithm (GA) • Genetic Algorithm (GA)+ANN Low Efficiency Low Accuracy 16 • Model Free • Offline • Manufacturing line has to be stopped • Cannot deal with variations, including part location errors, part geometry errors and environmental errors. Robotics Laboratory Ingram School of Engineering Bring Ideas Together Single Work Station Optimization • First Time Through (FTT) rate • How to balance FTT rate and cycle time? – – – – FTT rate is calculated statistically. Cycle time is recorded each assembly Modeling? Optimization methods? • FTT prediction? Cycle time estimation? 17 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Multi Work Station Optimization • FTT rate of the whole system • Cycle time of the production line 18 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Performance Optimization • Performance optimization – Knowledge transfer • What can we transfer? How to transfer? n1k n 2k n kM e1k e 2k e kM y1k y 2k y kM x1k x 2k x kM u1k u 2k u kM Stage 1 Stage 2 Stage M Multi-Stage Learning Methodology 19 Robotics Laboratory Ingram School of Engineering Bring Ideas Together Discussions • Comments – Robot teaching – System optimization • Interesting topics? – 20 Robotics Laboratory Ingram School of Engineering Bring Ideas Together