Bulldozers are frequently used to execute work in hazardous environments. Teleoperation provides a viable method for allowing operators to perform work without being directly exposed to the hazards of these environments. However, removing the operator from the bulldozer presents many challenges associated with reduced task engagement and controllability. The relocation of the operator and placement within a remote teleoperation system may be simplistically modelled as placing a filter on the operator's inputs and outputs. The various cues that would excite the sensory systems of an operator on board the bulldozer must be remotely replicated. However, the fidelity and timeliness of cues provided to the teleoperator are unavoidably constrained by limitations in the systems that effect this replication. Likewise, there are similar implications for accurately achieving a machine response in accord with the command outputs of the operator.This thesis focuses principally on the inputs to the teleoperator with the aim of identifying how perception enhancements might be applied to alter the characteristics of this input filter. The motivation for this research is to determine what factors are critical to achieving high levels of teleoperation performance and user acceptance.To conduct this investigation, an enhanced perception cell capable of high fidelity replication of motion, visual and aural cues is integrated with an existing bulldozer teleoperation system. The cell enables targeted analysis of the influence of individual feedback cues on performance and user acceptance. Experiments have been conducted with the enhanced perception cell for a structured bulldozing task. Results indicate that visual quality is a dominant factor influencing operator performance. Motion feedback provides no additional benefit beyond that provided by enhanced visual quality. The value of task visualisation to support accuracy and planning is also highlighted.