Figure 9.

Sensorimotor control framework. (a) Ui and Sj are steps i and j of the ULP and SLP, respectively, M is a column vector with the motor signals, is a column vector with the derivatives of the x and y positions taken at every time step, P_M and P_S are physical processes involved in producing muscle forces and extracting sensor information (respectively), Q is a 2 × 4 matrix (each sensor against each motor signal) containing the self-organized motor synergies, S* is a column vector with the goal information (i.e., the difference between the target position and the current position of the pendulum in 2D), α is a column vector containing the modulations of the goal parameters, ϕ contains the modulated goal parameters, ⊗ means elementwise multiplication, S′ is a column vector containing the difference between the final and the initial position of the pendulum at every iteration of the SLP, and e is the error signal between S′ and S*, which is used to modify the α parameters at every iteration). (b) The connectivity matrix Q obtained (the matrix has been transposed for visualization purposes). (c) A schematic of the pendulum platform with four muscles M_1–4 and the x and y axes from a camera placed under the pendulum. (d) The reaching error obtained as a function of the SLP iteration number (see text).