RA RACE CE A softw A softwar are e fr framew amewor ork k - PowerPoint PPT Presentation

RA RACE CE A softw A softwar are e fr framew amewor ork k for inter or interope operable, ble, plug-and plug and-play play, , distributed r distributed robotic obotic sys systems tems-of of-sys systems tems Robert Skilton t Skilton |

Experienc Experience e fr from JET om JET – The w he wor orlds lds most most po powerful nuc erful nuclear fusion lear fusion experiment xperiment • >30,000 hours of operational experience • 8 maintenance programmes • >20 Years of robotic maintenance • >2000 tools used • >8500 components handled Replacement of the entire “first wall”, using over 350 separate remote handling tools and replacing over 7,000 components over 18 months. 2 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Challenges Challenges for lar or large ge-scale r scale robotic obotic maintenance a maintenance and decommiss nd decommissioning ioning • Integrating (100s of) systems from multiple suppliers • Multiple bespoke interfaces • Training requirements • Operational lifetime may be several decades • Requirements evolve over time • Obsolescence management • Scalability • Reliability 3 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Uses of Uses of softw softwar are in these r e in these robotic obotic facilities acilities Hardware interface - hardware integration and translation Control – control methods ​, generic control laws, motion planning, AI Sensor processing – signal processing, sensor fusion, feature extraction HMI & GUI – integrated user interfaces to provide operator control and interaction Planning & scheduling tools - which can be used to add process and procedure to control tasks. Virtual & augmented reality - to display information in the form of intuitive 3D visualisations / digital twin. Condition monitoring - to track patterns and changing conditions in the equipment in order to identify developing faults. 4 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Design Design requir equirements ements Modularity – Control should be implemented in discrete blocks, so that sections can be removed, swapped out, and used in conjunction with other discrete blocks. Standardisation – Data should use pre-defined, but extendable, data structures. This allows for customisation while keeping as much compatibility as possible. Extensibility – Control should be implemented in such a way as to allow it to be extended by later applications. For example, a 4-wheeled robot could be extended to control a 6-wheeled robot. Reusability – Control should be implemented in a generic fashion. For example, a 4-wheeled robot is controlled as a 4-wheeled robot, rather than brand specific model. Common, user interfaces – A single, universal, consistent interface for users and developers. 5 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Object Object-Oriented Design? Oriented Design? • Encapsulation • Abstraction • Inheritance & Composition • Polymorphism Not about the software, but about the system connected by the software. OO alone does not create interoperability or plug-and-play capability. Introspection and knowledge of a type hierarchy allows us to go further. 6 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Wha hat is t is Cor CorteX teX? Software Framework CorteX attempts to solve the main problems associated with interoperable, plug-and-play, Data Representation distributed robot systems-of-systems, at least from a data/communications perspective. Communications CorteX could be thought of as: 1. A standardised graphical data representation for robotic systems 2. A method for communicating this representation 3. A software framework that implements the above Tools 4. Additional software tools to add functionality 7 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Self-descript Self description to allo ion to allow a sing w a single le standardised standa dised interf interface ace • CorteX uses a self-describing, standardised data representation. Data Representation • Every node (Simplex) in the graph uses the same format. • This data representation can be used as part of a communications protocol to allow distributed components of a single control system to exchange data without prior knowledge of each other. • This means a CorteX control system can grow to incorporate new hardware and control features without modifying other distributed components.​ 8 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Da Data R ta Repr epresenta esentation tion Simplex Simplex (graph nodes) Data • Type (from type hierarchy) Type • ID ID • Data (name, value pairs) • Relationships (graph edges) Data • Commands (definition of command and parameters) Relationships Commands (Definition) 9 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Type Hier ype Hierar archy hy imple imple rocessor anipulator oncept u stem I rocessor oor inator erial arallel A is artesian igital I ive orwar Inverse oncept rocessor ripper Arm ontroller anipulator anipulator anipulator oncept oncept oncept an set inematics inematics oor inator elector u a R Ro oti inger Ro oti inger Ro oti 3 inger inear A is Rotar A is u a Inverse A is artesian oncept II A ripper ripper ripper oncept oncept inematics ontroller ontroller elector Rotar A is ol A is inear A is artesian elta ontroller ontroller ontroller ontroller Types Data Rules Relationship Rules Command Rules 10 10 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Type Hier ype Hierar archy hy – Manipula Manipulator E tor Example xample imple imple rocessor anipulator oncept u stem I rocessor oor inator erial arallel A is artesian igital I ive orwar Inverse oncept rocessor ripper Arm ontroller anipulator anipulator anipulator oncept oncept oncept an set inematics inematics oor inator elector u a R Ro oti inger Ro oti inger Ro oti 3 inger inear A is Rotar A is u a Inverse A is artesian oncept II A ripper ripper ripper oncept oncept inematics ontroller ontroller elector Rotar A is ol A is inear A is artesian elta ontroller ontroller ontroller ontroller 11 11 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Str Structur uctural al & Mor & Morphological phological Rules ules Behavioral Elements Structural Elements 12 12 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Example System Ar Example System Architectur hitecture Arm u a R Ro oti inger II A ripper Rotar A is ol A is Rotar A is inear A is inear A is inear A is oncept ontroller oncept oncept ontroller oncept orwar rocessor oncept Rotar A is inematics elector elector oncept artesian artesian elta artesian u a Inverse Rotar A is oncept ontroller oncept inematics oncept artesian igital I oncept oncept ive an set 13 13 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Wha hat tools ha t tools have been de e been developed, using eloped, using CorteX Cor teX? Examples: • Hardware interface • Control • GUI • Planning • VR / AR • Condition monitoring • AI / Autonomy • Etc. 14 14 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Model-driv Model driven, en, automa automaticall tically gener generated ted GUIs GUI • Byproduct of self-description and standardized data representation. • GUI only needs compiling, unit testing once • Human Factors, Accessibility standards, look and feel consistent across whole facility • Reduces training need • Just one example of interoperable behavior and benefits 15 15 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

16 16 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019

Cur Current & ent & p planned deplo lanned deployments yments • 2x Integrated Innovation in Nuclear Decommissioning project teams • 3x InnovateUK projects on maintenance, repair, decommissioning • Joint European Torus remote maintenance systems • European Spallation Source Active Cells Facility handling systems This software has the potential to reduce the cost to develop, integrate, and maintain robotic facilities in high-consequence environments. 17 17 | RoboSoft: Software Engineering for Robotics Royal Academy of Engineering, London, UK, 13-14 November, 2019