Introduction Modeling of the robotic hand prosthesis’ finger Finger prototype test platform set-up Morphology optimization Results Conclusions and Perspectives Morphological optimization of prosthesis’ finger for precision grasping of little objects J. L. Ramírez 1 , A. Rubiano 1 , N. Jouandeau 2 L. Gallimard 1 , O. Polit 1 1 LEME Université Paris Ouest Nanterre La Défense, France { jl.ramirez_arias, astrid.rubiano, laurent.gallimard, olivier.polit } @u-paris10.fr 2 LIASD, Université Paris 8, France n@ai.univ-paris8.fr July 2015 1/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Finger prototype test platform set-up Morphology optimization Results Conclusions and Perspectives Plan 1 Introduction 2 Modeling of the robotic hand prosthesis’ finger Description of the robotic hand prosthesis’ finger Kinematic model Dynamic model 3 Finger prototype test platform set-up Materials and methods Kinematic tracking and force measure 4 Morphology optimization Evolution process Evaluation process Experiment 5 Results 6 Conclusions and Perspectives 2/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Finger prototype test platform set-up Morphology optimization Results Conclusions and Perspectives Soft Robotics Classical Robots Rigid structures Soft Robots [Nurzaman et al., 2013] Elastic and deformable bodies Unconventional materials [Andrianesis and Tzes, 2013] Improve interactions with the environment UB-HAND IV [Palli et al., 2012; Ficuciello et al., 2014] Pisa-IIT Soft Hand [Ajoudani et al., 2013] 3/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Finger prototype test platform set-up Morphology optimization Results Conclusions and Perspectives Robot Features Tendon driven mechanisms Flexible links Smooth joints Morphological analysis [Jouandeau and Hugel 2013-2014] To reach better synergies between movement primitives and limbs lengths Applied to NAO humanoids To validate parts dimension on real To design primitives in simulation 4/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Description of the robotic hand prosthesis’ finger Finger prototype test platform set-up Kinematic model Morphology optimization Dynamic model Results Conclusions and Perspectives Tendon-driven finger composed of three joints: Metacarpophalangeal (MP or MCP) - θ 33 Proximal interphalangeal (PIP) - θ 35 Distal interphalangeal (DIP) - θ 36 Under-actuated ⇒ one servo motor ⇒ angle joints relations: θ 35 = 0 . 23 θ 33 θ 36 = 0 . 72 θ 33 Fastening point Flexion - Extension DIP Extension tendon PIP Pulley Flexion tendon Servo Motor Fastening point MP Up - Down 5/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Description of the robotic hand prosthesis’ finger Finger prototype test platform set-up Kinematic model Morphology optimization Dynamic model Results Conclusions and Perspectives Denhavit-Hartenberg - Khalil and Kleinfinger (DHKK) 𝒚 𝒈 Link α i a i d i θ i 𝒜 𝒈 33 − π / 2 0 0 θ 33 𝒎 𝟒𝟓 𝜾 𝟒𝟕 34 π / 2 0 0 θ 34 𝒚 𝟒𝟕 𝒜 𝟒𝟕 𝒛 𝒈 35 − π / 2 l 32 0 θ 35 𝒚 𝟒𝟔 𝜾 𝟒𝟔 36 0 l 33 0 θ 36 𝒛 𝟒𝟕 𝒎 𝟒𝟒 𝒜 𝟒𝟔 f 0 l 34 0 0 𝒚 𝟒𝟓 𝒛 𝟒𝟔 𝒚 𝟒𝟒 𝒎 𝟒𝟑 𝜾 𝟒𝟒 � � n 0 R n 0 P n 𝜾 𝟒𝟓 0 T n = i − 1 T i = ∏ 𝒛 𝟒𝟒 𝒜 𝟒𝟓 0 0 0 1 𝒜 𝟒𝟒 i = 1 𝒛 𝟒𝟓 6/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Description of the robotic hand prosthesis’ finger Finger prototype test platform set-up Kinematic model Morphology optimization Dynamic model Results Conclusions and Perspectives Virtual displacements and virtual works Virt. Disp. of q Virt. Works Forces Q T δ W e = δ r e e M ¨ q T δ W i = δ r i 𝜾 𝟒𝟕 𝜾 𝟒𝟔 𝒚 𝟒𝟕 𝒚 𝟒𝟔 ⇓ 𝒚 𝟒𝟒 𝒙 𝟒𝟓 Dynamic equilibrium 𝒙 𝟒𝟒 𝜾 𝟒𝟒 𝒛 𝟒𝟕 𝒛 𝟒𝟔 𝒈 𝑺 δ q T [ M ¨ q − Q e ] = 0 𝒛 𝟒𝟒 𝒙 𝟒𝟑 ⇓ Input torque τ 33 ( f R , q , ˙ q , ¨ q ) 7/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Finger prototype test platform set-up Materials and methods Morphology optimization Kinematic tracking and force measure Results Conclusions and Perspectives The experiments seek to: Track the kinematics = ⇒ CCD camera Prosilica GE-2040 1 Measure fingertip force = ⇒ Flexiforce � Sensor 2 Evaluate tendon driven dynamic = ⇒ Using different motors 3 (classical and serial actuactors, from 2 . 3Kg-cm to 101Kg-cm, from Traxxas to Dynamixel) Interchangeable Actuator Adjustable finger position 8/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Finger prototype test platform set-up Materials and methods Morphology optimization Kinematic tracking and force measure Results Conclusions and Perspectives Finger position overshoots The 0 P x f vector shows perturbations after contact Sample experiment with Traxxas actuator (2 . 3Kg-cm): The lengths of the finger could: Increase the amount of torque needed Impact the precision of the grasping 9/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Evolution process Finger prototype test platform set-up Evaluation process Morphology optimization Experiment Results Conclusions and Perspectives ⇒ Reach a constant f R of 5N Find ⇐ Min position error Optimal finger’s phalanges lengths Min input torque τ 33 Morphological Optimization Evolution process based on an heuristic evaluation Simulation of Kinematic of our finger Simulation of Dynamics of our finger 10/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Evolution process Finger prototype test platform set-up Evaluation process Morphology optimization Experiment Results Conclusions and Perspectives Motors, Torques < M , T > and lengths new as parameter values Algorithm 1 evolution < M , T > ( n , H , eval ) 1: ( H , L ) ← ( / 0 , / 0 ); 2: for i = 0 to n do 3: lengths new ← newParam < M , T > ( H ); 4: ( d , m ) ← move ( lengths new , q initial , q obj , U , dt ); 5: score ← eval ( d , m ); 6: if score == ACCEPT then 7: insert (( lengths new , score ), L ); 8: end if 9: insert (( lengths new , score ), H ); 10: end for 11: return best ( L ); 11/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Evolution process Finger prototype test platform set-up Evaluation process Morphology optimization Experiment Results Conclusions and Perspectives Evaluation of positioning error d and input torque m ( i.e. τ 33 ) Algorithm 2 eval ( d , m ) 1: if d < d best then 2: ( d best , m best ) ← ( d , m ); 3: return ACCEPT ; 4: else if m ≥ 0 then 5: if m < m best then 6: ( d best , m best ) ← ( d , m ); 7: return ACCEPT ; 8: end if 9: end if 10: return REJECT ; 12/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Evolution process Finger prototype test platform set-up Evaluation process Morphology optimization Experiment Results Conclusions and Perspectives Algorithm 3 kinematicMove ( lengths new , q initial , q obj , U , dt ) 1: q ← q initial ; 2: t ← 0; 3: while contact ( q ) == false do ( u , t ) ← next ( U , t , dt ); 4: q ← f ( lengths new , q , u , dt ); 5: 6: end while 7: return ( dist ( q , q obj ), − 1); 13/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Introduction Modeling of the robotic hand prosthesis’ finger Evolution process Finger prototype test platform set-up Evaluation process Morphology optimization Experiment Results Conclusions and Perspectives Algorithm 4 DynamicMove ( lengths new , x initial , q obj , f R , U , dt ) 1: q ← position ( x initial ); 2: x ← x initial ; 3: t ← 0; 4: while contact ( q ) == false do 5: ( u , t ) ← next ( U , t , dt ); 6: x ← g ( lengths new , x , u , dt ); 7: q ← position ( x ); 8: end while 9: while torque ( x ) < f R do 10: ( u , t ) ← next ( U , t , dt ); 11: x ← g ( lengths new , x , u , dt ); 12: end while 13: return ( dist ( q , q obj ), u ); 14/18 Nicolas Jouandeau Morphological optimization of prosthesis’ finger
Recommend
More recommend
