You are on page 1of 20

PhD Thesis Structure

Karna Potwar
How to write a thesis
• Describe a situation
• Describe the problem that arises from the situation
• Describe how others have approached the problem
• Explain the need to approach in a different way or expand upon what
has been done
• Say what you aim to do
• Number: +919421994640
Initiating something obstacle
• Why do i not start?
• Before starting somethingi associate the outcome to negative
• If something expected is negative that means input has no value
• I need new input
• This arises due to thinking into the future
• Shall i override the expectation and work?
• It is only an expectation that is whyi do trial and error
• If something is certain that means I am god till that event arrives and then i am god for all of them.
• But that might happen or not
• If it happens then bad
• How do i decrease the probability of something negative happening?
• By increasing the probability of something positive happening.
• How do i increase positive probability
• By doing things that have proven to provide good results
• In my example storify my thesis
How to storify
• Connected series of events:
• Thesis:
• Read papers:
• Papers have hypohesis
• What is hypothesis? Researcher thinks of doing something and expecting something in his mind
• Tries that ‚doing‘ through a model : gets a different result
• Hypothesis fails
• But that is just one input and expected out put does not arrive
• Ry other inputs:
• Some of them closer to expected output
• Then they talk why is it closer to putput and why others arent
• Describe their findings
• Different paper : has other hypothesis different input, different utput or same output
• Based on such papers, new researcher trieshis own input and expected output
• Basically, thesis is about trying to prove what others failed in or trying your own hypothesis and proving it
• Based on what your own hypothesis comes up?
• Basically every hypothesis comes up through reading papers
• My hypothesis comes up through a certain article suggesting foot rotates faster for heeled gait.
• Gradiet is steeper at TD and LO with increased instability
• COP as hypothesis to find heeled gait, not just shift COM but have a dynamic profile.
• COP stroke , shift COM but stance leg conrol?
• Basically for t1 and t2 the force vector will be different because change theta will be different.
My story
• Motivation :
• Why gait, movement, task completion, lifestyle, healthy, prevent musculyr atrophy?
• Nothing can be achieved by stadnstill?
• Movement to do tasks, move based on hope
• Action required to achieve something
• For survival humans have to eat food and have relationships
• To eat food, humans require money
• Money is earned by doing work
• Wor can be sedentary or requires movement
• In either case humans have to move from a point to point in any world, to not have muscular atrophy
• Hence, locomotion is important
• Locomotion can be done barefoot but in todays age with the advent of technologa and comfort, we have
footwear that allows human to prolong locomotion
• Similarly, foot orthotics help in improving lifestyle of people with movement disorder.
• Why heeled footwear
Why heeled footwear
Discuss points :
• Baerfoot not on all terrain, then we got shoes shoes did not show Forced COP
any clinical relevance but only statistical significance. movement
COP timing with
• Constrained movement and pathology consrained movement
vertical and rotation.
• Freedom and constrained gait can be affected Increasing solutions by
• Less power more power parameter additions
• Sync no sync How do these
parameter can be
• Pathology and heeled constrained movement, explore similarity. used? Towards forced
• Constrained movement timing.
• By understanding properties of heeled gait, orhotses design in
pathological gait can be facilitated
• It is a vast field and no overlap between the two has been
established
• We test at the surface foot interface
Why heeled footwear
• Because : foot dynamics are dependent on external forces and force tranmission occurs at COP
• By altering COP position we monitor force on the COM
• Accuracy in COM movement by using COP progression, ensure accuracy in force direction
• COP variability but a pilot study to see effectiveness of COP
• COP can have different progressions, but to understand transition from barefoot to heeled gait, we approach with a hypothesis.
• In typically barefoot S curve, but heeled is unstable faster foot rotation to achieve forefoot contact, thereby dd/dtheta is greater
for heeled gait.
• Use a heuristic function to transition from barefoot to heeled gait and then corroborate GRF and spatio-tmeporal parameters
• GRF,
• Indi param : to see if all of them lie within model predictions, and how does mdel param change (show closest value for each
case and (max error)
• Multi param : already done
• Show increase in solutions
• Balance, rotation, vertical and horizontal displacement should be in sync
• By using COP progression we control the push off timing based on forced COP movement through convex inserts
Previous prediction
• Talk about heeled gait characteristics in comparison to barefoot
• How are these used in conjuntion with stroke gait
• We first extract required params COP GRF, SS DS sl, d
• Then we extract speed based correlation because heeled gait speed changed, SS, DS sl changed, (take correlations of all these)
• See for similarity ? None
• No gait deerminants
• Stroke has some aspects, sway and speed,
• For sway back and forth in AP. Speed control,
• Test sway COP=COM, test estimation COM-COP is gait based, slow speed VBLA?, just to fit GRF, can very high stiffness help?, it does
• Use linear COP profile similar to heeled gait, Not use tangent because we do not know how was the progression from barefoot to
stroke
• We straight away test linear profiles. By using linear profile we induce similar COP progression as heeled gait., can aler COP
progression through convex insers that change locus.
• We see how much does it estimate:
• Speed is an issue.
• Medium speed estimation through numeric optim. Not search every K. Can be done but a lot of parameters two legs
• To test similarity, we try to use COP data to estimate stroke gait
Sroke additional result Brainstorm:
GRF fit

• Points to add
• Heel and stroke gait similarity on COP curve
• TD angle condition choices.
• Stiffness choice (check Farley, Lee, Kuo paper)
• GRF estimation IMU insole papers, compare with your model
Motivation
• Foot ankle dynamics is affected by pathological gait and footwear. The direct
impact may be observed on the COP progression. Factors such as plantarflexion,
foot orientation and foot placement will have a cumulative effect on COP.
• Can a COP progression model simulate effects of footwear and spastic muscles?
• Model with postural control have been used to simulate pathological gait or
healthy gait, Can COP progression be responsible for postural control?
• In a way we can replace foot – ankle links withs simple AP based COP
progression. Foot-ankle joints require at least parameters such as, toe angle,
ankle joint angle, foot actuation/passive spring and foot length
• If we can replace them with COP displacement and COP acceleration, we would
reduce simulation time and simplify the model.
COP progression for heeled footwear
• Barefoot wlaking
• Attributes – repeat
• Template IP and SLIP results
• COM trajectory, COM and GRF trajectory
• IP is like walking on pegs with no ankle joint.
• IP+COP : ankle joint, COP displacement (variable)
• COP model
• COM-COP inclination angle wiht COP displacement
• What does it look like in SLIP +COP
• Profile shifts up or down but qualitatively doesnt change
• Heel gait, faster foot snap (hypothesis)
• What curves represent this? Show all possible curves.
• Select a handful for hypothesis testing (as it fits with tan function)
• Derive equations
• Explain update function with and without acceleration.
• Algorith for SLIP+COP
• Choosing error minimization function (other candidates, in discussion)
• How does pofile look like? COP for different gamma
• FP distributino across d and gamma. Stable.
Contd.
• Barefoot walking GRF :
• time scale, different speeds, not as good as Jung probably because we did not try all parameters
• Similar pattern;
• One can just provide velocity minimization to obtain different starting COP velocities.
• Increaseing stiffness similar to Jung.
• COmbine with literature
• Heeled gait :
• Tried a different approach, checked with individual curves for ss, ds, cad, strln,
• Ss and ds fit best suggest 2 of the determinants are comparable with high gamma value.
• Aim was to use a cumulative effect of ST error reduction and GRF, but individual gait determinant values.
• COmbine with literature
• Spatio-temporal distriution for , ss,ds,str, cad
• Effects of d and gamma on output values and model parameters.
• Combine with literature
• CUmulative effect
• Barefoot and heeled walking,
• Combien with literature
• General model behavior
• Assumptions
• Limiations
• COP curves
• Error minimizatino funciton
• Obtaining COP data, insole, forceplate
• Obtaining shoe stiffness and model stiffness
• Using manipulaive COP methods to relieve foot pressure?
Stroke Gait
Premise from heeled gait : decreased push off and lesser joint movement. Similar to stroke gait. But we obtain
speeds as low as 0.9 m/s with previous COP model, By not considering control and actuation, can stroke gait
kinetics and ST be modeled using SLIP with an alternative COP model?
Aim to achieve slower speeds, estimate ST and stroke gait kinetics.
ST can be improved with COP progression(heel gait), can it also estimate GRF forces?

Hypothesis 1: Slow walking speeds not modeled by SLIP because it cannot model low froude number gait
features.
Can COP progression eradicate this aspect?
COP progression is regulated by plantar flexion – Winter
During quiet standing COP position provides restoring force.
SLow walking similar postural control , ?as observed by hof in dynamic stability
- Candidate 1 : COP acceleration = COM acceleration – redirects GRF? Wha basis? Nolan footdrop COP.
Candidate 2 : COP acceleration observed in stroke gait affects GRF? Linear fit.
- Literature on GRF modeling with bipedal models VBLA and FMCH. Used
Stroke gait
• Data collection and preprocessing?
• Extracted GRF, COP and ST param for both leg combination.
• Relation between ST parameters. And walking speed.
• Extracting ST param. How? MP-LP
• Why? SLIP has k, theta, v_avg : Can any parameters be described a s a function of
walking speed to reduce parameter search?
• Temporal parameter correlation ? St, ds, ss
• No strong correlation observed
• Extract FL curves, to model stiffness, shows no stiffness relation
• Extract COP in AP direction with theta hypothesis 2.
Bipedal model
• SLIP similar description as heeled gait.
• Asymmetry : explain different properties, kr kl, thetr thetl : Alternative apexes are similar as Merker
paper.
• Except we induce asymmetry to find periodic solution instead of symmetry and perturbation.
• Two apex states, TD and LO conditions are similar
• COP progression
• Cand1 : COP = COM Nolan condition
• Xcp--=xcom.. : IP model for standing explaing stability condition of standing,
• Slow walking : propagation + stabilization : to check if COP relation holds here and VBLA and FMCH showed decreased valley of GRF,
• To investigate if stability is provided and GRF valley is flatter
• Cand2: COP and theta relation : to check similarity with heeled gait. Similar linear curves.
• Algorithm : SLIP with asymmetric pivot
• Instead of fixing E, fix v_apex: Hence, E=mv2+mgy+k(1-y)2;
• Do not search as we were not looking for solution distriution more towards limited search of v_apex
• V_apex and y fixed and theta, obtain k and x_cp:
• Error minimiztaion function
Results
• Check for GRF for cand 1 and cand 2 best fit
• Check for indi param, and cumulative search for cand 1 and cand 2 to
check effect on spatio-tempo and GRF/Power estimation cumulative
effect
• Recommendation results?
• Discussion:
• Ind err – one leg, two leg
• Why only spatio-temporal parameters, classify stroke gait
• GRF and power, Kinetic modeling towards soft orthotic design based on
apssive stiffness.
Conclusion
Discussion
• Aim and general results
• Model architecture
• COP profile
• Having an autonomous system
• Stability of models
• Validation with footwear
• Importance of stiffness
• Importance of COP acceleration
• Asymmetry handling
• Subjective and generalization models
• Limitations
• Future study
• Conclusion
• Insights gained
• Improvement of previous models
• Future endeavours
• 4.03 : Complete figures for stroke
• 5.03 : Go through results for heled gait
• 6.03 : brain storm stroke discussion points and write
• 7.03 : Check Matteo's thesis tense, think of a story.
• 8.03 : read thesis and edit
• 9.03 : read thesis and edit
• 10.03 : read thesis and edit
• 11.03 : read thesis and edit
• 12.03: read thesis and edit
• 13.03: read thesis and edit
• 14.03 : read thesis and edit

You might also like