Assessment of the reliability of output data for automotive technical examination of a road traffic accidents

Abstract

This paper addresses the reliability of input data used in forensic traffic accident reconstruction and shows that uncertainty in key parameters (road– tire friction coefficient µ, braking distance/mark length s, time intervals, video measurement scale and frame rate, and electronic event records) is a primary source of instability in expert conclusions. The aim is to develop a practice-oriented framework to (1) formalize measurement and interpretation errors, (2) quantify their effect on typical reconstruction outputs (initial speed, braking and stopping distance, time-to-collision, and the technical possibility to avoid the crash), and (3) report results as intervals or probability distributions rather than single point estimates. The proposed workflow combines a deterministic physical braking model with probabilistic uncertainty propagation. Input parameters are represented by justified ranges or distributions and propagated through the model using Monte Carlo simulation to obtain distributions of output quantities and relevant quantiles. A sensitivity analysis step is used to rank the most influential inputs and to justify which measurements should be prioritized in on-scene documentation. In addition, the framework includes cross-validation of heterogeneous evidence sources (dashcam/CCTV video and EDR/ACM data) by analyzing the residual speed difference ∆v(t) and diagnosing systematic biases caused by scale calibration, variable FPS, or time synchronization errors. A demonstration braking case study confirms that uncertainty in µ typically dominates the overall spread of the estimated speed, while the measurement error of s often provides a secondary contribution. Therefore, improving reconstruction reliability primarily requires better characterization of road surface conditions/friction and rigorous documentation of braking evidence, complemented by consistent digital-data checks. The results support the development of uncertainty “passports”, standardized checklists for data acquisition, and more transparent and reproducible forensic calculations.