Ref. Ares(2016)2533586-01/06/2016 H2020 RIA COMANOID H2020-RIA-645097 Deliverable D4.1: Demonstrator specification report M6
D4.1 H2020-RIA-645097 COMANOID M6 Project acronym: Project full title: COMANOID Multi-Contact Collaborative Humanoids in Aircraft Manufacturing Work Package: WP 4 Document number: D4.1 Document title: Demonstrator specification report Version: 1.0 Delivery date: Nature: Dissemination level: Authors: M6 Report Public Adolfo Suarez (Airbus Group) The research leading to these results has received funding from the European Union H2020 Program/2015-2020 under grant agreement n o 645097 COMANOID. 2
Contents Executive Summary... 4 1 Introduction... 5 2 Demonstration Environments... 6 2.1 A350 Aircraft... 6 2.2 A350 Physical Mock-up... 6 2.3 A320 Physical Mock-up... 6 2.4 Simulator (V-REP)... 6 3 Demonstration Scenarios... 7 3.1 First demonstration scenario... 7 3.2 Second demonstration scenario... 7 3.3 Final demonstration scenario... 8 References... 9 3
Executive Summary The present document D4.1 Demonstrator specification report is part of Task 4 : Integration, demonstration and benchmarking. The document describes a set of iterative demonstration scenarios that are expected in order validate the project achievements and guide the future work. As stated on the project proposal, 3 main demonstrations are expected: - At M24 Individual technological bricks and a first integration are validated - At M36 Different task sequences are validated individually on a simplified environment - At M48 Fully autonomous mission execution is tested on a realistic environment. The document also describes the associated demonstration environments and tasks. This deliverable is complemented by several confidential annexes including a set of simplified 3D data of the demonstration scenarios in various formats and more detailed descriptions of the environments, tasks and challenges. Confidentiality: The content of this document, the images and the attached documents and 3D models are the property of Airbus and Airbus Group. Their use is granted for project purposes only. Any publication of the descriptions or images contained in these documents or of any media generated using the provided 3D models is subject to specific written agreement of Airbus Group. 4
1 Introduction COMANOID investigates the deployment of robotic solutions in well-identified Airbus airliner assembly operations that are laborious or tedious for human workers and for which access is impossible for wheeled or rail-ported robotic platforms. This document describes the use cases that will integrate and validate the technical developments of the project. One of the target situations is illustrated in Figure 1. The task takes place in the cargo area of an aircraft being manufactured. The worker is located on the side of the plane, and has to access items located on difficult to reach areas on the side of the plane. In addition, as can be seen the area is crowded with rigid and flexible systems (electrical harnesses and pipes), which need to be handled with care. Figure 1 Assembly/Inspection operation in the cargo triangle zone Work in this zone is often considered as non-ergonomic. For a humanoid robot, the access requires necessarily to use multi-contact strategies to exploit the surroundings in order to support itself, ensure stability and increase its operational forces. In terms of perception, it is required to build and interpret dense maps of the environment in order to localize itself, infer the nature of the objects and unambiguously detect the structural elements which can be used as force relays. A major advantage of the proposed environment is that a realistic 3D mock-up is available and can be used to perform the different project demonstrations without affecting the production planning and taking risks on actual airplanes. At the end of the project actual tests on a real aircraft can be performed if the maturity of the solutions is validated by Airbus and all major risks for the humanoid, the aircraft and the workers is excluded. In addition of the physical mock-up, a virtual environment has been created and will be provided to partners as a set of both COLLADA and V-REP files. This environment has been geometrically simplified in order to cope with the capabilities of current simulation environments. This was done by reducing both the number of objects and by simplifying the existing meshes. 5
2 Demonstration Environments Four demonstration environments are proposed (Figure 2) and are described in this section. A350 Real Aircrafts A350 Physical Mockup A320 Physical Mockup Simulation (V-REP) Figure 2: Complete and Simplified virtual environments 2.1 A350 Aircraft Demonstrations could be performed on the real aircraft before the end of the project, but his is not a requirement for the project. Because entering an actual aircraft requires specific approvals that will only be granted by Airbus if the maturity proven by the COMANOID technology developments is considered sufficient. Previous validations in simulation and laboratory environments will be necessary. 2.2 A350 Physical Mock-up A physical mock-up of the A350 section is available for the project. This scale 1 mock-up is representative of the actual airplane section. However, a significant number of parts were machined from metal blocks instead of being assembled from production parts. Given that the A350 has a significant number of composite parts, the materials and textures might differ from the aircraft even if the geometry is very close. The big advantage of the physical mock-up is the fact that it can be used for the project without minor time constraints and that it can be slightly adapted where needed to fit the project needs, while remaining very representative. 2.3 A320 Physical Mock-up The physical mock-up of the A320 section is made out of actual parts build for the A320 stress tests. The structure is therefore weakened. A section was cut from the rest of the aircraft and is for the moment awaiting to be prepared for use as a real mock-up. The A320 is smaller than the A350 and so will be the physical mock-up. The height of the cargo area in the A320 is also smaller and was a potential issue for the project. The floor of the main section (passengers) was not recovered and will not be part of the mock-up. 2.4 Simulator (V-REP) A complete simulation environment has been prepared by Airbus Group. It contains the 3D model of the A350 physical mock-up as well as the access scaffolding. The Virtual Robot Experimentation Platform (V-REP) is a portable and flexible simulation framework allowing rapid algorithm development, system verification, rapid prototyping and deployment for cases such as safety/remote monitoring, training and education, hardware control, and factory automation simulation [1]. V-REP is developped by Coppelia Robotics [2] and uses internal scripts in lua[3] but can also call external code via plugins and is readily connected with the ROS environment [4]. 6
3 Demonstration Scenarios The proposed demonstration scenarios are guidelines for the project lifespan. Three demonstration scripts with increasing difficulty are presented. The goal is to progressively validate individual technologies and to iteratively integrate them on scenarios where the challenge and the complexity of the tasks are well adapted. The proposed scenarios shall be reviewed by the consortium before and after each demonstration in order to align their difficulty to the intermediate project achievements while keeping a demanding set of goals and high level of excellence. 3.1 First demonstration scenario The goals at M24 are to validate individual bricks and their first integration on tasks that do not modify the environment (moving around, mapping, inspecting and pointing). Demonstrations should mainly take place on the A350 physical mock-up with simplified environment. Two parallel streams of demonstration are expected: Validating all perception technical bricks developed in WP2 on the COBOT platform in the environment of demonstration: Localization and map generation with dense 3D data (T2.1) Object recognition (T2.2) Visual tracking with 2D and 3D sparse data (T2.3) Contact skin and associated safety strategies (T2.4) Validating preliminary multi-contact planning and control humanoid algorithms developed in WP1 and integrating preliminary safety constraints in the partners lab environment but with near-realistic physical mock-up. If possible, a first test visit to an actual aircraft will be also organized. The purpose is to test the algorithms maturity on the real environment; to identify improvement axis and to gather a first 2D and 3D map of the environment. 3.2 Second demonstration scenario The goals at M36 are to validate simple tasks with reversible impact on the environment (fitting a glued bracket, removing a floor tile) and to test complex postures in a complex environment. Demonstrations should take place both on the A350 physical mock-up and if the results are conclusive on the A350 aircraft. Two kind of benchmarking will be performed in the Airbus Saint- Nazaire environment: Demonstrating the ability of humanoids to navigate in the aeronautic environment, inside and outside of the aircraft. The path of the robot will be free of obstacles (human and non-human) but will contain most of the static difficulties, that is: stair, aircraft cargo door to span, non-flat ground of the cargo area (frame and stringers), etc. The robot will be already equipped of the perception system developed in WP2 and validated @M24 on the COBOT. Performing elementary process tasks in static postures inside the aircraft structure demonstrator or outside the aircraft. The static positions may be complex but pre-programmed off-line. These demonstrations will be performed both on HRP-4 and TORO robots. If possible, a visit to an actual aircraft will be also organized. The purpose is to test the project results at this stage and to anticipate the final demonstration scenario and issues. 7
3.3 Final demonstration scenario The goal at M48 is to perform the entire scenario in the demonstration environment in order to validate the project results in a task as realistic as possible, validating the complete autonomy and reliability of the solutions. Demonstrations should take place both on the A350 physical mock-up and if the results are conclusive on the A350 aircraft. The displacement of the robot will be fully automated, mastered by the mission manager provided and adapted by AIRBUS, including transition phases between navigation and posture reaching. The mission would certainly be refined, but the partners agreed on this rough sketch: The robot starts at floor 0 The robot walks and climbs stairs to reach the aircraft demonstration area The robot grasps the parts or tools disposed at approximately known positions on top of a table (printing machine, torqueing tool, brackets) The robot finds the entrance of the aircraft cargo area (by using localization with respect to a continually updated 3D map of its environment) The robot enters into the aircraft (using multi-contact planning/motion and/or robust walking) The robot moves into a complex area (using multi-contact planning/motion) The robot performs one or several predefined tasks with high accuracy relative to the aircraft parts The robot changes the area and its position and performs again the same or different task The robot goes out of the aircraft The robot returns to floor 0 8
References [1] ROHMER, Eric, SINGH, Surya PN, et FREESE, Marc. V-REP: A versatile and scalable robot simulation framework. In Intelligent Robots and Systems (IROS), 2013 IEEE/RSJ International Conference on. IEEE, 2013. p. 1321-1326. [2] V-REP simulator: http://www.coppeliarobotics.com [3] Lua: http://www.lua.org [4] M. Quigley, B.Gerkeyy, K. Conleyy, J. Fausty, T. Footey, J. Leibsz, E.Bergery, R. Wheelery, A. Ng. ROS : an open source Robot Operating System. In Proc of IEEE Int. Conf. of Robotics and Automation, Kobe, Japan, May2009. 9