VIEWS: 9 PAGES: 32 POSTED ON: 12/3/2011
Neuromechatronics: Frog lab - neither neural nor mechatronic Goal: use muscles to control movement 1) implant intramuscular electrodes 2) stimulate implanted muscle pairs 3) measure the movement 4) alter stimulation to control the direction of movement FES: functional electrical stimulation - to restore function after paralysis - in conjunction with a neural prosthesis - use neural signal to control a paralyzed limb Frog lab - muscle properties - state dependent torque motor - non-linear activation function - frog hindlimb anatomy - experimental set up and A/D - animal handling and dissection Muscle properties - alpha motor neuron and all the muscle fibers it innervates a motor neuron muscle A motor pool B motor pool A tendon aponeurosis muscle B Mechanisms of force production - force length properties FMO 0.5 LMO LMO 1.5 LMO force produced by the muscle is highly state dependent - non-linear function of muscle length Mechanisms of force production - force velocity properties lengthening Empirical fit: vM = b(FMO – FM) / (FM + a) shortening force power power velocity => muscle force nonlinear function of velocity Mechanisms of force production - modelling muscle properties Hill-type models: ‘active state’ F = A(t)*F(l)*F(v) force no interactions between activation, length, and velocity effects, linear in activation (all inaccurate) velocity length Activation dynamics - from neural input to ‘active state’ of muscle from neural input to muscle activation to muscle force depends on calcium dynamics e.g. first order activation dynamics dA(t)/dt + [1/tact (b + (1 – b)u(t)] a(t) = u(t)/tact Tendons tendon confuses relation between anatomy and muscle length musculotendon system LT1 LMT LMT LT LM - tendon compliance LM a(t) muscle contraction FM = FT dynamics LT2 vM vMT - tendon compliance vT LT = LT1 + LT2 => we need to model tendon compliance (DFT/DLT) Interaction dynamics between muscle and tendon LT1 LMT musculotendon system LM LMT LT - tendon LM compliance a(t) FM = F T muscle contraction LT2 dynamics vM vMT tendon - compliance vT LT = LT1 + LT2 LMT a(t) FT musculotendon skeleton system vMT Muscle action from its anatomy - muscle causes movement by acting through its attachments LMT a(t) FT musculotendon skeleton system vMT hip semitendinosus predicted ST path knee 1cm 1cm 0.2N ankle Muscle properties - strong state dependence of force on length and velocity of movement - activation dynamics, between neural input and ‘active’ muscle contraction - tendon dynamics mediate interaction between muscle and limb - limb movement from muscle contraction via muscle anatomy Non-linear effect of activation strength - activation strength as reflected in stimulation frequency … varying stim rate at a constant length … varying length at a constant rate => nonlinear scaling of FL curve with activation strength Non-linear effect of activation strength force length curve is altered at different strengths Non-linear effect of activation strength - modelling the nonlinear dependence Model activation dependence predicted FL curves on stim rate (f) and muscle length everything else is a free parameter Non-linear effect of activation strength - effects on force-velocity function FV curve at different stim rates FV curve at different lengths Non-linear effect of activation strength predicted FV curves - modelling the nonlinear dependence FV dependence on length FV dependence on delayed activation and length - introduce ‘effective length’ as delayed memory of length Virtual muscle (Cheng Brown and Loeb) - empirically based model, but looking more carefully at interactions - but too complicated for the simple control here Muscle properties - strong state dependence of force on length and velocity of movement - activation dynamics, between neural input and ‘active’ muscle contraction - tendon dynamics mediate interaction between muscle and limb - limb movement from muscle contraction via muscle anatomy - non-linear effects of activation strength on evoked force => not a trivial motor to control (especially in motion) Frog anatomy - muscles of the dorsal thigh VI + RA VE BF SM vastus internus vastus externus biceps femoris semimembranosus rectus anticus action: action: action: action: knee extensor, knee extensor knee flexor, hip extensor, hip flexor hip flexor Knee flexor Frog anatomy - muscles of the dorsal thigh IP deep muscle, in between VE and BF iliopsoas action: hip flexor Frog anatomy - actions of muscles - evoked isometric forces BF IP Frog anatomy - actions of muscles - evoked isometric forces VE VI Frog anatomy - actions of muscles - evoked isometric forces RA SM Frog hindlimb muscles - muscles with complex variations in actions across the workspace => choose muscle combinations which allow a range of motion Experimental setup and A/D 1) Implant intramuscular electrodes - bipolar stimulating electrodes - generally nerve stimulation is better, but harder electrode configuration electrodes placed orthogonal to the orientation of the muscle fibers - create a voltage across a set of fibers (actually probably nerve) exposed region of electrode Experimental setup and A/D 2) stimulate implanted muscle pairs stimulation isolation unit - to protect the animal from outside power sources - output is proportional to the input biphasic stimulation to balance applied charge and reduce damage biphasic current pulse positive and negative phases same amplitude so there’s no net charge accumulation and no damage Use train of stimulation: frequency, amplitude, pulse width, number of pulses to specify response strength (specified in AO out) Experimental setup and A/D caveat: - we really don’t need anything greater than 1ms pulses to stimulate muscle - DDA AO card digitizes (under Matlab) at 500Hz => 2ms pulses mininum - long pulse durations, even with charge balancing, can cause damage if responses fade with repeated stimulation, might be possible to switch to the sound card instead Experimental setup and A/D 3) measure movement Two ways: video tracking (today) - use webcam to track movements of the leg following stimulation - calculate direction of movement from video - this is relatively straightforward technically and is most functional - but it’s the most difficult - limb dynamics come into play - state dependence of muscle actions isometric force (next class) - attach leg to force transducer - measure evoked isometric forces - calculate direction of force - more straightforward since limb dynamics and state dependence are irrelevant - (but transducers came in yesterday) Experimental setup and A/D 4) alter stimulation to control movement After measuring movement direction, update stimulation parameters to produced desired direction. offline: - apply stim train with one set of parameters - based on evoked direction of movement, change parameters - repeat until desired direction online - continuously monitor movement - alter stimulation online to affect movement Control parameters are up to you: - amplitude, pulse width (though not with 2ms min), frequency, number of pulses Experimental setup and A/D Today’s goals 1) make electrodes 2) dissect frog and implant muscles 3) set up A/D for stimulation 4) record video of leg movement to stimulation - for single muscles - for a pair of muscles, varying strength 5) choose 2 control parameters and compare effects - for a single muscle Lab report 4) record video of leg movement to stimulation - for single muscles - for a pair of muscles, varying strength 5) choose 2 control parameters and compare effects - for a single muscle - Quantify the above data - plots of stimulating two muscles, varying relative strengths - plots of effects of two different stimulation parameters - which is the best parameter for control?
Pages to are hidden for
"Muscles"Please download to view full document