Andrew Kwasniak, PhD, PE, PTOE, ACTAR

AI-Based AADT & Traffic Volume Estimation

DecisionLine applies trained AI models to traffic-volume estimation, including vehicular, bicycle, and pedestrian activity. The approach can combine available counts, roadway characteristics, probe information, land-use context, traffic-control characteristics, and other relevant inputs to estimate traffic activity where complete field-count data are unavailable.

The objective is not simply to generate a number, but to develop transparent and testable estimates that transportation professionals can evaluate within the context of the available data.

AI Crash-Report Analysis

Crash reports contain valuable engineering information, but much of it is stored in narrative text, diagrams, coded fields, and other formats that traditionally require substantial manual review.

Andrew's work includes the development of AI-assisted methods for extracting, organizing, comparing, and interpreting crash-report information so that source documents can be transformed into structured inputs for engineering and safety analysis.

From Crash Documentation to Engineering Analysis

One area of Andrew Kwasniak's AI work involves automated and AI-assisted accident reconstruction based on crash reports and available supporting information.

The objective is to move beyond document summarization. Relevant crash information is identified, organized, compared, and incorporated into a structured engineering workflow that can assist with reconstruction of vehicle movements, collision sequences, roadway conditions, and other relevant factors.

The resulting analysis is intended to remain reviewable by engineers rather than operating as an unexplained black-box conclusion.

Intelligent Error Detection & Autocorrection

Transportation and crash datasets frequently contain incomplete, inconsistent, or conflicting information. An AI system that simply accepts every source value can reproduce those errors in the final analysis.

DecisionLine's approach includes intelligent evaluation of input information for inconsistencies and potential errors. Where information conflicts, the system can compare related evidence and engineering constraints, identify the discrepancy, and support correction or reconciliation of the input before subsequent analysis.

This combination of AI interpretation, engineering logic, validation, and error correction is intended to produce more reliable and auditable analytical workflows.

Safety Performance Functions

AI and machine-learning methods can complement traditional Safety Performance Function analysis by helping transportation professionals examine complex relationships among roadway characteristics, exposure, crash history, and other variables while maintaining an engineering framework for interpreting the results.

Network Safety Analysis

Large roadway networks generate substantial amounts of crash, roadway, traffic, and operational information. AI-assisted analysis can help organize and evaluate those datasets, identify patterns, support network screening, and focus engineering resources on locations and conditions warranting further investigation.

Engineering Decision Support

The goal is not to replace engineering judgment. DecisionLine develops AI as a decision-support capability in which results can be tested, reviewed, compared with source information, and incorporated into defensible engineering decisions.

DecisionLine.ai

Artificial intelligence, machine learning, data optimization, engineering analysis, and decision-support applications for transportation and infrastructure.

SPF Tool

Transportation safety technology for Safety Performance Function analysis, predictive safety methods, network screening, and data-driven roadway safety.

Tatum Group

Andrew Kwasniak's broader professional background in transportation safety engineering, forensic engineering, roadway design, crash analysis, and human factors.