Background:
Gait
How Lucy walked
Laetoli footprintsResearch methods:
Reverse-engineering gaits
The input model
Simulation
Analysis
ApplicationsField work
Background:
Gait
Traditional methods for interpreting the functional significance of the fossil remains of early human ancestors rely on analogy: if a given fossil bone resembles the same bone in a living species, then it is assumed the behaviour of the extinct species resembles that of the living one.
However, there is just one living habitual striding biped: ourselves, but several species of early human ancestor are known to have coexisted in the past, each with their own unique skeletal characteristics And since the rare event of fossilization has preserved them, we can assume each was successful in its unique adaptation
Instead of trying to mix ‘n match morphologies and adaptations drawn from several living animals, better to work out what the whole extinct animal was best adapted to do
In industry, ‘reverse engineering’ techniques are used to determine function from morphology, using computer modelling techniques. WHY NOT APPLY THESE TO FOSSILS?
The most complete fossil skeleton of an early human ancestor is still that of ‘Lucy’, from 3.16 MYA
Were her species, Australopithecus afarensis, anatomically committed bipeds, or do some chimp-like features, such as long forelimbs, short trunk and curved digits suggest their bipedal walking would have been‘bent-hip, bent-knee’ like that of modern common chimpanzees?
Sources of evidence for how Lucy walked
The first source of evidence for how Lucy might have walked is obviously her skeleton, more specifically her leg bones and their joints. Here we see Lucy's knee (middle) with a modern human (left) and a chimpanzee (right). There is no single living analogue matching all the details of her skeletal morphology.
Laetoli foot prints
Another important piece of information available to us is the famous Laetoli footprint trail.
There is still considerable debate as to who or what made the foot prints, at the time of their discovery the only culprit available was Australopithecus afarensis, however subsequent fossil discoveries have presented science with too many 'usual suspects' for their authorship. Further detail of the Laetoli trail and our current research can be found at: Bipedalism or Bipedalisms? Mechanical Function of the Early Hominid Foot.
Research methods:
Reverse-engineering gaits
The primary technique used is dynamic modelling. This is a three stage process, firstly we construct an input model, then this model is run through a simulator and finally the results are analysed.
The input model
This requires two components, inertial properties and kinematics. Inertial properties are the basic physical characteristics of the body segments, segment length, centre of mass etc. Kinematics are the motion characteristics of the body, joint position, joint rotations etc. These properties can be "mixed and matched" between different animals:
Simulation
The constructed input model is run through a dynamic analysis package. For rigid segment models this can be done using a package such as Adams. However we have recently moved to using Madymo which allows us to incorporate FEA surfaces into the segment model to investigate surface contacts.
Here we have the results of two simulations a model with the estimated inertial properties of "Lucy" were given kinematics from humans walking with bent hips and knees BHBK (right) and a normal erect gait (left).
A simple test of whether a given gait will work with a given set of limb proportions is to see if the simulation can walk. From above human gait either erect or BHBK will work, using chimpanzee kinematics, when applied to the model failed completely. This does not however give an indication of the efficiency of the gait.
In these models, a crucial element is an ability to optimize multiple gait parameters. We shall use a technique from cybernetics: the genetic algorithm, which simulates the action of natural selection.
Forward dynamic computer simulations use the forces generated by the muscles to drive the model. These forces can be set by hand but walking never manages to proceed further than the first stride (Movie 1). However we can use an automated optimisation procedure based on Genetic Algorithms to improve the muscle activation pattern and this produces stable walking (Movie 2). This model can then be used for further analysis of the mechanics and energetics of bipedal locomotion.
This is an inverse dynamics musculo-skeletal model. Measurements are taken from physical skeletal material and used to construct the model. Kinetic and kinematic data from experiments using human subjects is then applied to the model allowing us to estimate muscle forces and velocities.
Here we have another set of simulations this time using MADYMO software produced by TNO Automotive. Originally developed for building virtual crash test dummies for the motor industry, this system allows us to construct hybrid models consisting of rigid segment elements, which allow for rapid computation and FEA components, which are compute-intensive, but allow accurate modelling of surface contact parameters (such as friction) for studying foot contact parameters.
This project is studying the functional significance of the morphology of the foot in humans and the other great apes. Models were constructed using measured segment proportions and mass properties of human, chimpanzee, gorilla and orang-utan. The models are driven by real motion captured from the real animals. Simulated ground reaction forces produced by the models can then be compared with actual measurements obtained using a force platform. This allows us to validate the model's integrity and then to apply the procedure to extinct forms, and so predict ground reaction forces based on their morphology.
Analysis
These two graphs show the predicted power output from the "Lucy" simulations. Left, erect human bipedal gait; right, Bent Hip, Bent Knee (BHBK) human bipedalism of the type proposed for "Lucy" by some scientists. In BHBK walking energy is stored at the knee and ankle, predicting increased heat load.
Erect walking requires 1.2 W/kg* BUT BHBK walking requires 2.5 W/kg.*
* normalized absolute average power
Thus, the mechanical effectiveness of “Lucy” in BHBK walking would have been very low, but “she” could have been an effective upright walker But, does mechanical effectiveness imply energetic efficiency?
In studies of humans, emphatically: Yes! in human BHBK walking
- Oxygen consumption increases by 50%.
- Rate of perceived exertion increases by two units for the same speed.
- Heart rate shows an average 30% increase.
- Blood lactate production doubles during BHBK gait.
- And core body temperature shows significant rise, sustained well after cessation of exercise: sufficient to demand a recovery time in excess of 150% activity time: rise in heat load predicted by model is confirmed.
Our latest project will unite these mechanical and physiological approaches by development and verification of a model to predict the metabolic costs of gait from foot contact parameters.
Recording oxygen consumption and ground reaction forces during steady state walking.
3D plot of foot pressure in normal human gait.
Applications
- Biologically accurate models of the limb musculoskeletal system can be used to optimize/customize orthopaedic surgery and prosthesis design.
- Better understanding of role of gait parameters feeds back into virtual robotics and design of walking robots.
Field work
Members of PREMOG are conducting field work involving observational studies of locomotor and behavioral activities of primate species in various parts of the world.
Tarsius bancanus in Sabah, Borneo
