Friday, February 22, 2019

OMB Approves FRA Automation Induced Human Error Study ICR


Yesterday the OMB’s Office of Information and Regulatory Affairs announced that it had approved an information collection request (ICR) from the DOT’s Federal Railroad Administration for an Experimental Investigation of Automation-induced Human Error in the Locomotive Cab. The study will be conducted by the DOT’s  Volpe Center using their Cab Technology Integration Laboratory (CTIL).

The Study


The study will look at two different types of train automation systems currently in use by railroads in the United states; the Trip Optimizer and Electronic Train Management System (ETMS) Positive Train Control (PTC). According to the final Supporting Statement [.DOXC download] provided to OIRA, the study will assess three working hypotheses:

• Automation provides specific performance benefits (e.g., TO reduces fuel usage; PTC prevents overspeeding and transgressions into workzones or past a red signal) compared with manual control;
• Automation does not reduce perceived workload in the locomotive cab compared with manual control; and
Automation condition will show more errors in high workload situations than in low workload situations (e.g., distractions lead to failure to notice mode transitions) and the manual condition will not.

That Supporting Document provides a fairly detailed description of the proposed test. The idea behind this study is that disruptions to the engineer/conductor attention at critical junctures in train operation lead to errors. The specific disruption that will be studied will be a radio call from the dispatcher carefully timed to changes in operation of the automation system. The Supporting Document notes that an earlier study suggested that this might be a specific cause of operator error in using train automation systems.

The results of the study will be published as an FRA technical report at some future date.

Commentary


This looks like it will be an interesting study and it may have important implications for a number of other areas where automated safety critical systems require operator interactions.

The three hypotheses being tested here are an interesting look at automation systems in their own right. The first goes to the efficacy of the safety-critical automation system; if that assumption is not true, then the entire design of the system is called into question. The second hypothesis is a human factors issue, but it also is an important question of safety design. If the safety critical system requires operator action, it should not add to the operator’s workload else it increases the probability of a safety-critical human-error; that is the third hypothesis in a nutshell.

The FRA is fortunate that it has simulator capacity to evaluate these concerns. Designers of a process safety system at a chemical plant (for instance) are unlikely to have that capability.

No comments:

 
/* Use this with templates/template-twocol.html */