Extended Data Fig. 5: Sensitivity of simulated spherical treadmill rotation prediction accuracy during tethered walking to ERP and CFM constraint parameters.
From: NeuroMechFly, a neuromechanical model of adult Drosophila melanogaster

Spherical treadmill rotational velocities resulting from Kinematic Replay of walking depend on simulation constraint parameters. Shown are Spearman correlation coefficients computed between measured and estimated treadmill rotational velocities for (A) forward, (B) lateral, and (C) yaw axes when varying the simulation’s error reduction parameter (ERP), and the constraint force mixing (CFM). (D) The best combination of ERP and CFM—0.1 and 3, respectively (black outline)—was selected through a normalized weighted sum (NWS) of the correlation coefficients for each axis.