Systems & control Inspired by Biology Systems & control: challenges in the 21st century Richard van der Linde Delft University of Technology, Delft Biorobotic Laboratory & Altran Technologies Netherlands June 9, 2004 1
Contents Who are we? Introduction biorobotics Lessons learnt so far Conclusions Challenges for the 21 st century June 9, 2004 2
Who are we? The world TUD, Mechanical Engineering Man Machine Systems Prof. F.C.T van der Helm/Prof. P.A. Wieringa Delft Biorobotic Laboratory R.Q. van der Linde (ass. prof) M. Wisse (PhD student) D. Hobbelen (PhD student) G. Christiaanson (PhD student) A. Schiele (ESA PhD student) J. Frankenhuyzen (Eng.) E. Fritz (Eng.) June 9, 2004 3
What do we do? Biorobotics is robotics inspired by biological systems Key projects in the DBL: Efficient & stable walking 1995 STW Safe manipulator 1999 TUD Biocompatible grasping 2002 VENI June 9, 2004 4
Contents Who are we? Introduction biorobotics Lessons learnt so far Conclusions Challenges for the 21 st century June 9, 2004 5
A new generation of robotics is coming rehabilitation wheelchair robotics home robotics entertainment medical robotics agro robotics space robotics June 9, 2004 6
Interaction with biological systems requires a different control approach Biological inspiration = Biorobotics Useful properties of biological systems: stable interaction adaptive to environmental conditions safe interaction simple solutions Because we design for biological systems June 9, 2004 7
Human CNS control scheme NEURAL MUSCLE-SKELETAL force muscle intrinsic properties open loop neural input + - muscle dynamics force + + skeletal system position, speed feedback neural input muscle spindles gains F.C.T. van der Helm June 9, 2004 8
Also, biological and artificial systems have different system properties! human technique delay time 50-120 ms 0.1-5 ms # sensors hand 17.000 ~ 10-40 # sensors eye 12 Mp ~ 0.3-1 Mp # actuators 639 ~ 20-40 # DOFs ~ 200 ~ 20-40 # op. units 50-100 e 12 4-100 e 7 power cons. 100 W 4000 W Different solutions!!! June 9, 2004 9
Contents Who are we? Introduction biorobotics Lessons learnt so far Conclusions Challenges for the 21 st century June 9, 2004 10
Efficient & stable biped walking Problem: Energy consumption (price, complexity, non-natural) Honda ASIMO 43 kg 15 min. 26 DOF June 9, 2004 11
Efficient & stable biped walking Solution: Exploiting ballistic motion, creating stable limit cycles `1888 patent free swinging motions limit cycle analysis June 9, 2004 12
Efficient & stable biped walking gravity powered devices 2000 2001 M. Wisse 2002 June 9, 2004 13
Efficient & stable biped walking Problem: Activation & control Solution: phasic variation of passive stiffness -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 distance [m] g g Inman, 1981 γ June 9, 2004 14
Efficient & stable biped walking A simulation result: June 9, 2004 15
Efficient & stable biped walking 40 Problem: How to make efficient active springs Solution: Pneumatic (McKibben) muscles Force [N] 35 30 25 20 15 10 5 0 0 5 10 15 20 stretch [%] stretch force curve of McKibben muscle = June 9, 2004 16
Efficient & stable biped walking The first steps (2002) developed pneumatic components + PC104 controller Baps 3.4 kg 16 W (15 W elec.) 11 steps 3D June 9, 2004 17
Efficient & stable biped walking Problem: Achieving cycle robustness Solution 2: Mechanical Optimisation (spherical feet, DSF) (a) ϕ TL -0.2-0.4-0.6-0.8-1 -1.2 0.1 ϕ SL 0.2 0.3-1.5 no feet -1-0.5 (b) ϕ TL -0.2-0.4-0.6-0.8-1 -1.2 0.1 0.2 ϕ ϕsl SL -1 0.3 0.4-1.5 ϕ SL feet (R=0.5 L) -0.5 June 9, 2004 18
Efficient & stable biped walking Successful solution simple reflexes: stretching the knee swinging the hip Max 10 kg LEGO brick control steps 2D June 9, 2004 19
Efficient & stable biped walking 600 Problem: Achieving cycle stability in 3D Solution: Adaptation of reflexive control? 400 200 0-2500 -2000-1500 -1000-500 0 500 1000 1500 2000-200 -400-600 Limit cycle dead beat control June 9, 2004 20
Efficient & stable biped walking (muscle) powered devices 1995 2001? Van der Linde, Frankenhuyzen, Wisse, Schwab, Hobbelen 2002 2003 robust 3D stopping turning etc... June 9, 2004 21
A safe manipulator Problem: Arm movement in audience Solution: Passive compliant motion C1 C2 C2 C1 human movement control Lambda model (Feldman) June 9, 2004 22
A safe manipulator 4 DOF Gravity compensation mechanism of hand mass preloaded springs [Herder & Tuijthof, 1998] June 9, 2004 23
A safe manipulator 6 DOF passive compliant activation P1 = 5 [bar], P2 = 5 [bar], P3 = 5 [bar] 90 67.5 theta, ( -Mtheta ) 45 22.5 0 Optimisation of muscle configuration 45 67.5 90 112.5 135 phi, ( -Mphi ) June 9, 2004 24
A safe manipulator To be solved: dynamic position control June 9, 2004 25
Biocompatible grasping Problem: Instabilities during hard contact (e.g. hammer effect) soft objects hard objects June 9, 2004 26
Biocompatible grasping Solution: asymmetric teleoperation force feedback control approach ideal slave 10 Hz master C1 1000 Hz C2 June 9, 2004 27
Biocompatible grasping Experimental setup DC motor cable transmission adjustable damper finger interface position encoder adjustable leaf spring force sensor June 9, 2004 28
Biocompatible grasping stable for hard objects June 9, 2004 29
Contents Who are we? Introduction biorobotics Lessons learnt so far Conclusions Challenges for the 21 st century June 9, 2004 30
Conclusion In biological systems a plethora of control mechanisms seem to be present on different different functional levels, some of which are are active, some of which are passive. Being inspired by these mechanisms has resulted in biorobotic solutions that are simple, adaptive, stable & safe. June 9, 2004 31
Contents Who are we? Introduction biorobotics Lessons learnt so far Conclusions Challenges for the 21 st century June 9, 2004 32
The 21st century? Ǻström [2004] Control engineering has a soul but no body Van der Linde [2004] Biorobotics has a body but needs more soul dynamics biology design June 9, 2004 cybernetics electronics control 33