Force feedback is being used as an interface between humans and material handling equipment to provide an intuitive method to control large and bulky payloads. Powered actuation in the lift assist device compensates for the inertial characteristics of the manipulator and the payload to provide effortless control and handling of manufacturing parts, components, and assemblies. The use of these Intelligent Assist Devices (IAD) is being explored to prevent worker injury, enhance material handling performance, and increase productivity in the workplace. The IAD also provides the capability to shape and control motion in the workspace during routine operations. Virtual barriers can be developed to protect fixed objects in the workspace, and regions can be programmed that attract the work piece to a certain position and orientation. However, the robot is still under complete control of the human operator, with the trajectory being determined and commanded using the judgment of the operator to complete a given task. In many cases, the IAD is built in a configuration that may have singular points inside the workspace. These singularities can cause problems when the unstructured trajectory commands from the human cause interaction between the IAD and the virtual walls and fixtures at positions close to these singularities. The research presented here explores the stability effects of the interactions between the powered manipulator and the virtual surfaces when controlled by the operator. Because of the flexible nature of the human decisions determining the real time work piece paths, manipulator singularities that occur in conjunction with the virtual surfaces raise stability issues in the performance around these singularities. We examine these stability issues in the context of a particular IAD configuration, and present analytic results for the performance and stability of these systems in response to the real-time trajectory modification of the human operator.