This is the final project for Computational Motor Control (2020 EPFL course). We made use of the Central Pattern Generator model explained in class for simulating a salamander capable of both swimming and walking. What's even more important, we were sucessful in developing a smooth transition from walking to swimming thanks to a fine parameter tuning. We explain the totality of our results in this report.
The project is divided in several subexercises where we increase the model complexity step by step:
- First, we set all the parameters in
robot_parameters.pyand define the differential equations innetwork.pyso that we can runexercise_example.pywithout a problem and the behaviour is similar to the expected one in theory. - Next, in
exercise_8b.py, we run a first grid search in order to obtain the most efficient (high speed and low energy) values for the amplitude gradient and the phase lag when swimming. - In
exercise_8c.py, we study more in detail how does the amplitude gradient influence the speed and energy of the salamander. - In
exercise_8d.py, we experiment again with the paramaters to induce turning and backwards swimming - In
exercise_8f.py, we find the most optimal parameter values for a correct coordination between spine and limb joints - Finally, in
exercise_8d.py, we adapt the salamander drive depending on its position in order to obtain a smooth land to water transition.