3 & 4 septembre 2016 YVERDON-LES-BAINS Boissons - petite ...
Rahe.populus.ch West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains LaRA - Laboratory for...
Transcript of Rahe.populus.ch West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains LaRA - Laboratory for...
rahe.populus.ch
West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains
RH2-Y: LaRa andRH2-Y: LaRa and
LaRA - Laboratory for Robotics and AutomationJean-Daniel Dessimoz 11th June 2007
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA2
LaRA and Robocup at Home
Introduction
Service Robots
RH2-Y Robot
RH2-Y Team
Specific Tests & Open Challenge
Concept for the future
More information
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA3
Introduction
RoboCup@Home is a new league inside the RoboCup competitions (1st year officially - Atlanta ‘07; 2nd year effectively - re Bremen ‘06)
Focuses on real-world applications and human-machine interaction with autonomous robots.
The aim is to foster the development of useful robotic applications that can assist humans in everyday life.
RH2-Y is our project and team in the context of RoboCup@Home experience, season 2, i.e. for Atlanta 2007.
We intend, in a house environment, to
Participate in specific tests : « follow and guide a human », « manipulate » an object, « navigate » in the home, learn from a human (« copycat ») and possibly find out « who is who)
participate in the « open challenge », in Atlanta, USA.
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA4
Mobile Robots for Services
Some of LaRA activities aim at developing novel solutions for mobile and autonomous robots, cooperating with humans with the goal to assist them.
This is well in line with what is proposed in Robocup-at-Home initiative (RAH).
Our team has taken part in Robocup 2006 (RAH, Bremen , Germany) and is currently upgrading our robot (RH2-Y) for the next meeting and competition in July in Atlanta-USA (Robocup 2007)
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA5
RH2-Y Robot (1 of 4)
New (with respect to picture on right): arm, hand, FiveCo controllers,, wireless micro and headphone, loudspeaker
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA6
RH2-Y Robot (2 of 4)
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA7
RH2-Y Robot (3 of 4)
RH2-Y Robot inherits a lot of the functionnalities of our other autonomous robots. Its architecture is as follows: - a laptop runs the software, - a Beckhoff API manages input output
signals, - a TCP/IP AXIS camera provides low-
level vision to our robot, - a two axis motor controller ensures
adequate robot motion, - an Ethernet switch provides
communication between those elements.
Real-time interaction and supervision is programmed in a proprietary multiagent environment, named Piaget. Novel contributions for Robocup-at-Home application include the integration of vocal communication (analysis and synthesis), wireless remote control ("chat" mode), as well as map and trajectory management.
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA8
RH2-Y Robot (4 of 4)
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA9
RH2-Y Team
The members of the team are currently mainly the following ones: Pierre-François Gauthey, Arvind Singh Gautam,
Ajit Pratap Singh, Parul Dubey, Prof. Dr. Jean-Daniel Dessimoz , Samuel Jabès, Pierre Boucher, Michel Brisseaux (re. fig.)
A significant help has also been brought by Varuj « Ben » Sukhsevi and Veerapot« Mike » Kitcharoen, RH1-y team, other students, through their contributions to our ARY robotic platform, as well as by members of the the technical services.
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA10
Specific test: Follow and Guide a Human
Phase 1: K ~170’000 Lin; E~100’000Lin/s Min. requirements: K~400Lin, E~=400/5min~1 Lin/s
Phase 2: not specified
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA11
Specific test: Navigate
Phase 1: K ~150’000 Lin; E~75’000 Lin/s
Phase 2: not specified
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA12
Specific test: Manipulate
Phase 1: K ~50 Lin; E~50’000 Lin/s
Phase 2: not specified
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA13
Specific test: CopyCat
Phase 1: K ~16’000 Lin; E~100’000 Lin/s
Phase 2: not specified
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA14
Specific test: Who is Who?
Phase 1: K ~16’000 Lin; E~50’000 Lin/s Phase 2: not specified
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA15
Open Challenge - Go & Pick an Object (1 of 2)
Scenario: The user throws an object (re. playing with a dog)
The robot… detects and locates the object
Gets closer
Picks the object
Brings it back to starting position
The task requires : Vision/perception Locomotion/navigation Prehension/manipulation Communication Decision-making
Cognitic performance level: K ~16’000 Lin; E~100’000 Lin/s
Applicability Picking dropped objects for people having
reduced mobility
Encouraging light physical exercises and human-robot cooperation
Entertainment
Benchmarking and training for integrated capabilities
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA16
Open Challenge - Go & Pick an Object (2 of 2)
Cognitic performance level: K ~16’000 Lin; E~100’000 Lin/s
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA17
Concept for the future
Cognitics framework
Vocal dialogue
Agile strategic programming
Interactive guidance Conceptual improvements
Integrate past successful approaches Trajectory management, navigation paradigms
O3NEIDA approach (IEC 61’499) Robust holonomic basis 3D TOF distance vision
Cognitive agent
nin
nout *
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA18
Concept - Cognitics framework
Cognitics framework Cognitics is a domain which encompasses the science
and techniques of automated cognition
The MCS framework provides definitions and metrics for the quantitative assessment of cognitive entities
Expertise : E=K/∆t [lin/s]
« does it (right and) fast»
Cognitive agent
nin
nout *+time
Some consequences:
Work in small contexts
Use goal-oriented models
Systematically assess the quantities of cognitive entities and requirements
Beware of asymmetry in input-output constraints
Notice the importance of access operators
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA19
Concept - Agile strategic programming
Multiagents 100 ns granularity VAL-type
instructions Piaget environment Distributed control
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA20
Concept - Vocal dialogue
Ten word recognition
“Unlimited” speech synthesis
3 complementary agents
Word processing agent
Dialogue manager
Strategy enforcement agent
Interactive guidance possible (incl. current motion - stop, retry, cancel)
Beware of difficulties in normalisation of signal intensity and speaker specifics
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA21
Concept - Navigation paradigms
Deterministic
Probabilistic
Learning with modelling
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA22
Trajectories are typically defined as follows : - Select key points (x,y) on a map, - Interconnect them with linear sections - Resample the trajectory at regularly-spaced intervals - Smooth the trajectory, iteratively until some vehicle-related, maximum curvature constraint is met (reference for curvilinear smoothing) - Define local orientation (tangent angle) for each sample (x, y, theta) Trajectories are typically tracked as follows: - match vehicle position and orientation in terms of main structure - define relative position and orientation (transformation) of each wheel or limb (thereby superposing gait) - interpolation is made directly at individual wheel or limb level between so-defined successive positions along trajectory (reference for general trajectory definition and tracking):
Concept - Trajectory management
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA23
Concept - Robust holonomic basis (e.g.Festo)
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA24
Concept - 3D TOF distance vision (e.g. CSEM, IFM)
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA25
Concept - More advanced concepts yet
Improved controllers and elements (size, weight) Standards for virtual and real worlds (y c. URBI) Attitude, acceleration, world position sensors Locomotion with 4 articulated limbs Artificial skin Cultural engineering / robot education
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA26
Conclusion
Service Robots: assisting at home RH2-Y Robot: recent instance of ARY family robots RH2-Y Team: international (Canada, France, India, Switzerland ,Thailand) Four Specific Tests & Open Challenge Challenging concepts for next steps More information: re links
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA27
Acknowledgements
Many colleagues, engineers and students/alumni of iAi, as well as various network partners have contributed to or inspired our current solutions.
RH2-Y, West Switzerland Univ. of Appl. Sc., Yverdon-les-Bains, Robocup @ Home 2007, Atlanta-USA28
More information
Useful links : • More information on ARY and RH1-Y robots: descriptions and examples, pictures and videos : http://ary1.populus.ch, http://rahe.populus.ch • iAi : Industrial Automation Institute( "Institut d'Automatisation Industrielle »). Our team belongs to the LaRA (Laboratoire de Robotique et Automatisation) which is one section of the institute. • Robocup@Home : THE useful link, if you want to know more about this new league. The website is relevant about everything you need to know. • RoboCup2006: Dedicated to the "classic" RoboCup competition. Both of them (RoboCup & @Home) take place together (Bremen, Germany, June 2006). • Robocup2007: http://www.robocup-us.org/ (Atlanta, USA, July 2007)• (Swiss)Eurobot : a little bit off topic, but this competition is cool too. Each year some of our students participate to the challenge. • LaRA.heig-vd.ch : includes references and publications of our lab.