Operator interventions in autonomous off-road driving: effects of terrain

Abstract

-We pariicipaied in field trials of a semiautonomous vehicle. This gave us an apporiunity io coileci data on operaior inierveniions. In this paper we present an analysis of why and how operators intervene and examine the eflciency of these inierveniions. We compare the results from two trials in very dcferent ierrain io determine how this affected the time and success of the operaior interventions. In particular, we look ai the time it took operators to acquire situational mvareness in order to take the proper inieraciion. We discuss the implications of ihis on human-robot interaciions and inierJaces.

Common metrics for human-robot interaction

“…For example, the NASA-Task Load Index (NASA-TLX) [20], has been widely used to measure human performance and workload in teleoperation scenarios [8,25,44]. In general, results have shown that subjective ratings of workload decrease as the level of system autonomy increases and that shorter teleoperation tasks yield lower workload ratings.…”

“…In general, the most important aspect of using SAGAT to measure SA is performing a detailed task analysis in order to formulate appropriate operator queries. In Scholtz, Antonishek et al [44], for example, an analysis of human interventions in autonomous rover off-road driving was used to develop questions for assessing SA at multiple levels.…”

“…One well-known query-based tool for evaluating SA is the "Situation Awareness Global Assessment Technique" (SAGAT) [ll]. SAGAT has been used to assess SA at a various levels of autonomy [25,44]. In general, the most important aspect of using SAGAT to measure SA is performing a detailed task analysis in order to formulate appropriate operator queries.…”

Attention and Communication: Decision Scenarios for Teleoperating Robots

“…The research area of remote control of semi-autonomous land vehicles for search and rescue has received recent attention [9][10][11][12][13][14][15].…”

“…So, in many discussions of the control of semiautonomous vehicles, operators are expected to handle multiple vehicles (e.g. [13]); the number to be controlled is often mentioned as about four.…”