Despite many advantages, parallel manipulators are known to possess drive singularities where the control of one or more degrees of freedom is lost. Around these singular configurations, the required actuator forces grow unbounded. Previous efforts in the literature put forward singularity-consistent trajectory planning and singularity robust modification of the dynamic equations as a solution to this problem. However, this previous method is applicable only for the open-loop operation of the manipulator, whereas initial configuration errors, external disturbances, and modeling errors should necessarily be taken into account in a closed-loop sense in real-life applications. With this aim, a switching inverse dynamics controller is proposed in this study for the trajectory tracking control of parallel manipulators as they pass through drive singular configurations. Simulations of the application of the developed controller result in good tracking performance, even in the presence of modeling errors, while the actuator efforts remain bounded and continuous in the neighborhood of the singularity.