1 / 23

( ANguilliform Robot with ELectric Sense)

A new sense for underwater robotics in confined environments : the electric sense . GDR ROB GT2 (22 Novembre 2012 , Paris-Jussieu) Frédéric Boyer, Vincent Lebastard. ( ANguilliform Robot with ELectric Sense). 1. Aim of Angels.

asher
Télécharger la présentation

( ANguilliform Robot with ELectric Sense)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. A new sense for underwaterrobotics in confinedenvironments : the electricsense.GDR ROB GT2(22 Novembre 2012, Paris-Jussieu)Frédéric Boyer, Vincent Lebastard (ANguilliform Robot with ELectric Sense) 1

  2. Aim of Angels • To build a re-configurable anguilliform swimming robot… • Able to split into smaller robots (or modules) propelled by propellers • Where each robot: • can sense its environment (obstacles, objects and co-robots) and also communicate (co-robots) with a bio-inspired sense little known to roboticists: “the electric sense”. 2

  3. Objectives of Angels With these basic choices, we are exploring the potentialities offered by: • Self adaptation of the morphology to the situation of the perception (size and shape of the objects…) and locomotion • Multi-robot cooperation and communication … in order to design an AUV adapted to a wide spectrum of uses, in particular in environments where vision or sonar are of no use (murky waters with particles, confined environments, shallow waters…). 3

  4. What is electric sense? In order to perceive its environment, the fish polarizes two regions of its body… Gnathonemus petersii By comparing the currents crossing the skin with and without objects, the fish perceives the objects (shape, locations, electric colors…) 4

  5. The electric sensor Bio-inspired approach… African dipolar fish Physical principle: U imposed, I measured … based on this principle, we have built a set of slender probes… (ARMINES) [IEEE Sensor, sub.] 5

  6. Electrolocation test bed Object electrolocation Electric sensor test-bed ANGELS module (SSSA) + electric sensor (ARMINES) 6 dSPACE rapid control prototyping device

  7. Inverse electric problem Any electrolocation algorithm has to solve the inverse electric problem… Originally, it appeared as the most difficult theoretical one that Angels addresses. Basically, it can be stated as follows: 1°) First formulation: “Find such that in the scene : For Bc imposed and measured on the sensor.” Inverse problem Impedance – Tomography problem EEG algorithms From CVLab (EPFL)

  8. Inverse electric problem Too costly to be used on-line… Reduction of the parametric space constant on sub-domains Solvable with classical nonlinear control techniques… Considering simple shaped objects analytic solutions … 2°) Second formulation: “ Find and such that the electric matrix equation: For any and respectively imposed and measured.” Inverse problem Finite dimensional non linear problem

  9. Inverse electric problem We begun by adopting… • For navigation a Kalman filtering based approach [IJRR, to appear.]… • … gave first encouraging results but some limitations arose: • Requires an analytical model of the scene • Due to the sensor range, it requires to change the state dimension … Limitations became a serious drawback due to the increasing complexity of the project (complex scenes, multi-agent…). Action Perception Solution: Use a reactive approach! 9 In two steps…

  10. Exploitation of action-perception synergies • 1°) Deeper understanding of models [IEEE TRO 2012] based on: • Method of reflections: Interactions sensor-object = successive reflections , where: electric response produced by the sensor to recover its equilibrium currents reflected by the object with no sensor currents with no object • Exploitation of the morphology of the sensor: Bio-inspiration , Slenderness: Bilateral symmetry: Axial currents Lateral currents tells us if object is conductive or insulating, : if it is on left or right

  11. Exploitation of action-perception synergies 2°) Coming back to the nature… From IIBCE The idea consists in implementing this navigation strategy on the rigid modules… 11

  12. Exploitation of action-perception synergies In order to achieve this navigation strategy [IEEE TRO sub.]… • Remove the basal component from measurements • Apply the control law: The sensor aligns on the current lines. With simple , one can encode relevant navigation behaviors, as : • Seeking conducting objects • Avoiding insulating obstacles Method of potentials where Potentials are not virtual but real 12

  13. Exploitation of action-perception synergies Analytical study of stability Using the method of reflexions... Closed loop Kinematics: Repulsive law Attractive law ANGELS' Review 13 Reactivity increases with k/V

  14. Exploitation of action-perception synergies Simulations Behaviour = seeking conductors and avoiding insulators 14

  15. Exploitation of action-perception synergies Seeking a conductive object Portrait of the total electric field 15 Portrait of the perturbative component of the field

  16. Exploitation of action-perception synergies Experiments • Behaviour = Seeking conductors and avoiding insultors 16

  17. Exploitation of action-perception synergies • Seeking conductors and avoiding insultors: experiments Tank with an active dipole Tank with no object 17 Tank + 3 insulators + 1 conduct. Tank with 2 insulators

  18. Exploitation of action-perception synergies From memory-less to mnesic reactive control New perspective to the pb of underwater navigation in confined environments… • requires no model can be applied to Angels module (bilateral symmetry) • the electric field is an extension of the body of the robot (embodiment) • exploits the haptic modality of electric sense • Sensory-motor loops exploiting perception-action synergies (solve inverse pb by navigation) • Virtual “Real" potential field approach for underwater navigation • Memory – less reactive control • uses only two measurements (head) electrodes (binocular sensor) If we enrich the measurements with neck electrodes (depth)… • We can design mnesic reactive laws encoding more complex behaviors as “object exploration”…

  19. Mnesicreactivelaw : Experimentalresults Exploration of a smallobjects…

  20. Mnesicreactivelaw : Experimentalresults Exploration of a large objects 20

  21. Mnesicreactivelaw : Experimentalresults Exploration of a large object (wallfollowing)

  22. Angels demo

  23. Angels demo

More Related