Evaluating Optimization Algorithms for Human Gait Simulations
Are you interested in helping us improve our human gait simulations? We use these simulations to predict the effect of interventions like prostheses and exoskeletons, and to perform detailed biomechanical movement analysis from wearable sensor data. We are looking for a student to investigate the optimization algorithm that we use to generate the simulation.
It can be very time consuming to generate the simulations, especially when three dimensional human models are used. Furthermore, the algorithm that we currently use does not always find a realistic movement. Therefore, we aim to investigate if we can improve the performance by changing settings or using another optimization algorithm. The thesis will make use of an open-source toolbox that has been developed to create human gait simulations. (https://github.com/mad-lab-fau/BioMAC-Sim-Toolbox).
We are looking for a student with a strong foundation in optimization algorithms like IPOPT. It is not necessary to have a background in human gait and biomechanics, as the analysis will focus on the algorithm performance.
The tasks are as follows:
- Literature search into available open-source gradient-based optimization algorithms and internal algorithm settings
- Selection of suitable algorithms and internal settings
- Creation of experimental procedure to compare algorithms
- Analysis of algorithm performance based on experiments
More information can be found here: Proposal_OptimizationAlgorithm.