cobotics oberview
TRANSCRIPT
Cobotics Overview
Komal Kambale System Science and Industrial Engineering Department
State University of New York at Binghamton
Outline Introduction Why we need cobots? Cobotic systems
Characteristics Constraints Applications
Cobots Vs. robots Case study Review of leading cobots Ongoing research at peer universities Viable research options @BU Conclusion
Introduction
An apparatus and method for direct physical interaction between a person and a general purpose manipulator controlled by a computer
Cobot= collaborative + robot
Pioneers & Inventors
James E. Colgate
Michael A. Peshkin
Rodney Brooks
Esben Østergaard
Need
Mass production will require both men and machine Injuries caused to operators due to repetitive tasks Safety risks for operators working in closed contact with
robots
Taxonomy of assist devices
Shortfalls of Existing Solutions
Provides high precision and speed, but only on a large scale Safety as design feature is rarely available Lack of tactile sense solutions are often too standardized and not generic End users struggle to adapt with cutting edge technology Lack of trust from end users
Objectives of Cobot
Collaborate with humans without the dangers and limitations of other assisted devices
Cope up with different situations and tasks
Direct the motion of human operator or payload without the use of powerful actuators ABB Cobot FRIDA demo [3]
Components of Cobotic System
Cobot Coworker Environment of Workstation
Cobotic System Characteristic
Safe, by design Trained, not programmed Adaptive Flexible and re-deployable Easily integrated Affordable Extensible
Cobotic System Constraints
Slower Less powerful Payload High precision
Applications
CNC machining Quality inspection PCB handling Gluing and welding Machine tending Pick and place
Understands people and environment Safer compared to robots Flexible and easy to use Tasks are performed similar to human
way Can be trained by demonstration No/minimum integration required Affordable
Unaware of surroundings Potential danger to human safety High precision and repeatability Definite operations for task
completion are required Need expert programmers Integration is costly Expensive
Cobot Vs.
Robot
Case Study Human-robot collaboration in manufacturing: quantitative
evaluation of predictable, convergent joint action Problem statement: computing entropy rate in Markov chain to
assess the convergence of robot computational teaming model and coworkers mental model
Experiment setting: simple pick and place task Coworker: place screws in one of the three available
positions Cobot: drill each screw
Case Study Continued
Learning method Cross training Reinforcement leaning
Post hoc subjects which change in strategy or inconsistent execution
Case Study continued
Entropy rate is sensitive to coworkers strategy Using probabilistic distributions of coworkers next actions to
compute his presence at various locations in robots vicinity When coworker deviates from high probabilistic workspace
change motion parameters
ABB- Yumi
Initially built with keeping electronics industry in mind Application industry
Small part assembly Consumer products Toy industry Watch industry
Not suitable for Paint industry Food grade Clean room applications
ABB- YuMi Continued
Technical specifications Payload: 0.5 kg per arm Reach: 559 mm Accuracy: 0.02 mm Weight: 38 kg
Cost : $40,000 approx
Universal Robots-UR3 Primarily developed for small
format assembly tasks Application industry
Automotive Food and agriculture Furniture and equipment Electronics technology Metal machining Plastic and polymer Picking and placing cream bottles
in Johnson and Johnson Greece factory
Universal Robots-UR3 Continued
Technical specifications Payload: 3 kg Reach: 500 mm Repeatability: ±0.1 mm Weight: 11 kg
Cost : $23,000 approx
Baxter and Research Continued
Learning trajectory preferences for manipulators via iterative improvement
Explore computer vision, machine learning, and artificial intelligence by training Baxters to pour water into a moving jar, learn to cook by watching youTube, and work with other robots
Baxter and Research Continued
Assistive robots for blind travelers Accessible interfaces Assistive interaction Effective cooperation
Baxter and Research Continued
Closed-loop global motion planning for reactive execution of learned tasks
Baxter and Research Continued
Improving Soft pneumatic actuator fingers through integration of soft sensors, position and force control, and rigid fingernails
Baxter and Research Continued
Program Baxter in such a way that the robot can perform in a magic show on stage alongside Tempest What is the affect space for a human-
magician performance? How to improve on current robot
animation techniques? What are challenges and opportunities
when designing human-robot performances?
Viable Research Options @BU
Cobot + quality assurance Statitical analysis of particular
manufacturing task performed by Baxter
Adaptive learning Exploring cobot response to various
training techniques
Safety Determine weather or not a certain action
is safe on a human
Conclusion
Current research level at peer universities involves vision control study and human-robot interaction
Cobots provide opportunity to automate tasks that were previously automated
Baxter research robot provides unique opportunity to various universities for R & D