Towards Artificial ATRON Animals: Scalable Anatomy for Self-Reconfigurable Robots

Transcription

1 Towards Artificial ATRON Animals: Scalable Anatomy for Self-Reconfigurable Robots David J. Christensen, David Brandt & Kasper Støy Robotics: Science & Systems Workshop on Self-Reconfigurable Modular Robots August 2006 David Johan Christensen Ph.D-Student Adaptronics Group, Mærsk Institute University of Southern Denmark 1 of 28

2 Overview Why study Self-Reconfigurable robots? The ATRON system. Scaling the robot. Challenge of Scalable Functionality Scalable Anatomy for ATRON robots Discussion, Future work & Summary???????????? 2 of 28

3 Why Study Self-Reconfigurable Robots? 3 of 28

4 Purpose from Scientific Perspective? Understand, explain and recreate the characteristics of life!! CEBOT CATOM CONRO M-TRAN 3D-UNIT FRACTA POLYBOT MOLECULE SUPERBOT MOLECUBE CRYSTALLINE METAMORPHIC Many other. Robotic Species Robot Robot Biological Species Biological Organism Evolve Self-Sufficient Self-Contained Self-Maintained Autonomous Adapt & Learn Self-Assemble Self-Repair Self-Reproduce Intelligent & Conscious Robotic Modules Cells Cells Self-Organize Self-Assemble (Chemicals) Self-Reproduce (Division) 4 of 28

5 Purpose from Engineering Perspective? We want to build Better Robots! Why should it Modular? Multi-Purpose => Versatility (by clever design) Mass-Produce => Low Cost Extendable => Scalable Why should it be Self-Reconfigurable? Self-Repair => Reliable & Robust (from redundancy) Self-Assembly => Automated Production Adaptation => Increased Performance (both functional and morphological adaptation) 5 of 28

6 ATRON 6 of 28

7 Our Platform: The ATRON System Module Characteristics: Single degree of freedom 4 male, 4 female connectors Batteries (had power sharing) Computation, communication, Sensing: Distance, Tilt, Encoders 800 grams pr. module Manufactured: 100 modules 7 of 28

8 Versatility of the ATRON System Manipulation & Locomotion (Real Time) 8 of 28

9 Versatility of the ATRON System Self-Reconfiguration (8x Speed) 9 of 28

10 Scaling 10 of 28

11 Scaling the Robot Scaling down module size (cm to µm) Scaling up number of modules (tens to millions) Why? Closer to the cell metaphor. Improve engineering metrics (e.g. resolution of shape) 11 of 28

12 Scalability Challenges of Self-Reconfigurable Robots What are the Challenges? Hardware What to build? Self-Reconfiguration Morph between configurations. Functionality (THIS TALK!) Fast responding and functional robots. Many other challenges. 12 of 28

13 Challenge of Scalable Functionality 13 of 28

14 The Challenge of Scalable Functionality Number of Cells/Modules Speed of Response Complexity of Functionality?????? 14 of 28

15 What is the problem anyway? Self-Reconfiguration too slow: (for fast response) Slower with more modules (more moving) Faster with smaller modules (faster modules) Modules are locked rigidly: Modules are stuck in a global lattice. Actuation Forces does not add up. So scaling functionality is hard. Basically constant with volume. Functionality reduces to self-reconfiguration. Functionality do not transfer from a 10 to a 1000 module robot 15 of 28

16 How can we approach the problem? How to build Myriad-Module Robots? Functional & Fast-Responding Approach: Defining a Robot Anatomy Biological Inspired by Animal Anatomies Scalable Anatomically Structures Bone, joints, muscles, nerves, etc.. Scalable Anatomy Artificial Scalable Creature Anatomy Robotic Modules Computers, Actuators, Sensors, Gears 16 of 28

17 Where does this comes from? Hieratical Organization of cells Differentiation dependent on role in organism Animal Animal Robot Robot Organs Organs Tissues Tissues Cells Cells Anatomi Inspiration Cellular Inspiration Anatomically Anatomically Structures Structures Modules Modules 17 of 28

18 ATRON Anatomy 18 of 28

19 Moving from idea to system ATRON Anatomy ATRON-Nerve ATRON-Arteries ATRON-Bone ATRON-Joint ATRON-Muscle ATRON-Skin 19 of 28

20 ATRON Anatomy: Neurons and Arteries ATRON-Neuron Computation ~ Microprocessors Coordination ~ IR-Communication Sensing ~ Sensors ATRON-Artery Transportation of energy ~ Power sharing Notation 20 of 28

21 ATRON Anatomy: Bone ATRON-Bone Support Weight of Robot. Strong Connector. Lattice Interconnection. Scales in 3D. Density Yield Strength ATRON-Bone 720kg/m^ MPa* Bio-Bone 1900 kg/m^3 50MPa *Østergaard et al., Design of the ATRON lattice-based selfreconfigurable robot. Autonomous Robots, to appear, Femur Bone ATRON Bone 21 of 28

22 ATRON Anatomy: Hinge Joint ATRON-Joints Connection and relative rotation of bones. Hinge Joint Scales along rotational axes. Reversible to global lattice. Hinge Joint 22 of 28

23 ATRON-Joints ATRON Anatomy: Ball-Socket Joints Connection and relative rotation of bones. Ball-Socket Joint Scalable in 3D. Reversible to global lattice using muscles as anchors. Hip Joint Ball-Socket Joint 23 of 28

24 ATRON Anatomy: Muscle ATRON-Muscles Actuation of myriad-module robot Parallelizable to achieve scalability 24 of 28

25 ATRON Anatomy: Skin ATRON-Skin Purpose: e.g. protection from environment. ATRON Surfaces that can deform and stretch However: Rigid constraints on ATRON Full of Holes Two sheets patterns for skin 25 of 28

26 Discussion, Future work & Summary 26 of 28

27 Lesson Learned Will it ever work? Probably not with the ATRON alone Why not? Morphological differentiation of modules seems necessary: Functional differentiation are not enough. 210 different cell types in adult human What then? (Future work) Collect more experience from different systems: Which anatomical structures? How to combine anatomical structures? Design novel SR system based on scalable anatomy. 27 of 28

28 Summary Challenge of Scalable Functionality: Fast responsive functionality myriad-module robots Our Approach: Biological Inspired Scalable Anatomy. Anatomical Parts: Bone, Joint, Muscle, Skin, Nerve. Scalable Anatomy for ATRON System.???????????? 28 of 28