Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
An important topic in designing neuroprosthetic devices for animals or patients with spinal cord injury is to find the right brain regions with which to interface the device. In vertebrates, an interesting target could be the reticulospinal (RS) neurons, which play a central role in locomotor control. These brainstem cells convey the locomotor commands to the spinal locomotor circuits that in turn generate the complex patterns of muscle contractions underlying locomotor movements. The RS neurons receive direct input from the Mesencephalic Locomotor Region (MLR), which controls locomotor initiation, maintenance, and termination, as well as locomotor speed. In addition, RS neurons convey turning commands to the spinal cord. In the context of interfacing neural networks and robotic devices, we explored in the present study whether the activity of salamander RS neurons could be used to control off-line, but in real time, locomotor speed and direction of a salamander robot. Using a salamander semi-intact preparation, we first provide evidence that stimulation of the RS cells on the left or right side evokes ipsilateral body bending, a crucial parameter involved during turning. We then identified the RS activity corresponding to these steering commands using calcium (Ca2+) imaging of RS neurons in an isolated brain preparation. Then, using a salamander robot controlled by a spinal cord model, we used the ratio of RS Ca2+ signals on left and right sides to control locomotion direction by modulating body bending. Moreover, we show that the robot locomotion speed can be controlled based on the amplitude of the Ca2+ response of RS cells, which is controlled by MLR stimulation strength as recently demonstrated in salamanders.
, ,