This paper describes a step-by step process for the development of test procedures for pre-production driver assistance systems. The process begins with a detailed engineering description of system performance and utilizes a universal description of the causal factors and resulting crash types as the foundation for a detailed analysis of crash data. The process ends with a set of objectives test procedures that can be applied to pre-production driver assistance systems that address lane-keeping/road departure performance. The quantitative estimates were obtained from national crash databases, namely, 2004 General Estimating system (GES) and 2004 Fatality Analysis Reporting system (FARS). There were 10,945,000 vehicles involved in crashes in 2004, of which 1,114,000 and 977,000 vehicles were involved in multi-and single-vehicle lane-keeping/ road-departure type crashes, respectively. Other factors such as trafficway flow, alignment, curvature, and speed were also analyzed to determine appropriate test conditions.

The results provide separate test conditions for single-and multi-vehicle crashes. The tests for multivehicle crashes include testing vehicles traveling in both directions;​ same and in opposite directions. Tests for vehicles traveling in the same direction involve driving that simulates undivided multi-lane roads. Testing for vehicles traveling in opposite directions involves driving that simulates both straight and curved two-lane undivided roadways. Single-vehicle crashes involve one test that represents a curved two-lane undivided highway with a narrow shoulder and another that represents a multi-lane undivided highway with a shoulder having a parked vehicle. All tests involve a driver traveling at speeds between 30 and 50 mph.

This is the first application of the new crash-analysisbased process for developing test procedures. Additional challenges in performing the tests and using the results to estimate crash avoidance benefits are not discussed in detail in this paper