Error analysis model for dual-agent collaborative monocular ranging system based on first-order error propagation theory

Abstract

In unknown environments, a single monocular camera struggles to achieve absolute distance measurement due to the lack of scale information. This paper proposes a method for absolute distance estimation of arbitrary targets based on the collaboration of two mobile agents. The method takes two agents as the basic implementation unit and is applicable to multi-agent systems, where any two agents with communication and collaboration capabilities can perform ranging using the same strategy. For this dual-agent collaborative monocular vision ranging system, a systematic error model is established based on first-order error propagation theory and the worst case linear accumulation assumption, which traces the propagation of errors from pixel quantization to final distance estimation. The system ranging chain is decomposed into three serial stages: baseline calibration, viewing angle difference measurement, and triangulation distance calculation. Analytical expressions for the error in each stage are derived, and the relative error formula for the final distance is synthesized. The model reveals the key factors affecting ranging accuracy, providing a theoretical foundation for system design and parameter optimization.