mul layer behavioral mo on for complex robo c control
play

Mul$-Layer Behavioral Mo$on for Complex Robo$c Control with >10 - PowerPoint PPT Presentation

Mul$-Layer Behavioral Mo$on for Complex Robo$c Control with >10 DoF MR. KEVIN WAGNER & DR. JOHN WRIGHT -MILLERSVILLE UNIVERSITY- Need The future of robo+cs in the US Automate manufacturing processes Remain compe++ve in the


  1. Mul$-Layer Behavioral Mo$on for Complex Robo$c Control with >10 DoF MR. KEVIN WAGNER & DR. JOHN WRIGHT -MILLERSVILLE UNIVERSITY-

  2. Need The future of robo+cs in the US ◦ Automate manufacturing processes ◦ Remain compe++ve in the global marketplace Robo+cs play a key role in that automa+on Humanoid Robots have a host of capabili+es that may reveal poten+al new uses in industry ◦ Humanoids are not typically used in manufacturing or repeated processes ◦ Humanoids CAN be used to develop new tes+ng and training methods for use manufacturing

  3. Overview Prior work with NAO has focused on simple mo+on control and sensory percep+on. How do we develop advanced mo+on control with Timeline? The need for this type of control is most evident in industrial or humanoid robots ◦ Those which u+lize 10 or more DoF ◦ NAO plaKorm contains 25 DoF

  4. History of NAO Research at MU 2015-2016 Increased Mobility – Custom Mo+on using Timeline – Single Behavior/Layer Techniques 2013 ATMAE - Ramos & Wright ◦ Programming the NAO Robotic Humanoid with Object-Oriented Programming Methodology 2014 ATMAE - Wells, Kilbourne & Wright 2014-2015 Visual and Auditory Percep+on Classical AI Work ◦ Advanced Dynamic Mo+on Control and Object Begin Assistant Tour Guide Work - Tracking for Humanoid Robo+cs Sabba+cal 2017 ATMAE – Wagner & Wright ◦ Mul+-Layer Behavioral Mo+on for Complex Robo+c Control 2013-2014 Two Addi+onal NAO Purchased Soccer Mo+on Using Timeline Single Behavior/Layer Techniques 2012-2013 First NAO Purchased Reading / Object Recogni+on / Facial Recogni+on / Basic Mo+on

  5. Assistant Tour Guide Development (2014-2015) u Classical AI Technique – Decision Trees u Speech Recogni+on u Vision Recogni+on u And a licle bit of humor! hcps://www.youtube.com/watch?v=or_pN1ico9M&t=9s

  6. Assistant Tour Guide Development (2015) u Music Files & Mobility hcps://www.youtube.com/watch?v=lWMgYskFrOg hcps://www.youtube.com/watch?v=4YB9uKdqGGQ&t=3s

  7. Assistant Tour Guide Development (2016) u Tracking with Custom Mo+on (Single Layer/Behavior (Timeline) hcps://www.youtube.com/watch?v=vY1PkwXDSgg&t=1s

  8. Timeline Editor

  9. Demonstra$on of NAO hcps://www.youtube.com/watch?v=edae7ny7r90

  10. Advantages of Layered Mo$on Programming vs tradi$onal linear programming Ease and efficiency in training ◦ One operator may train a complex system ◦ Only one part is trained at a +me Ease and efficiency in program altera+ons ◦ The en+rety of the program need not be retrained ◦ Operator may change only one layer at a +me Complex behaviors are now possible

  11. Future Work NAO Scooter Tourguide ◦ Hallway naviga+on ◦ Line/Color tracking Scrip1ng Implementa1on ◦ Faster execu+on OpenCV Vision Processing ◦ Expanded vision processing capabili+es

  12. Thank You – Ques$ons or Comments Kevin Wagner Dr. John Wright Phone: (717) 725-1722 Email: John.Wright@Millersville.edu Email: klwagne1@Millersville.edu Site: hcp://sites.Millersville.edu/jwright

Recommend


More recommend