This paper presents an experimental investigation\r\nof human control of vehicles carried out on the basis of\r\ngeneral theories on human movement. The longitudinal and\r\nlateral accelerations are studied, and their relations with theories\r\nof motor optimality principles, such as minimum jerk,\r\nminimum variance, and the two-thirds power law are\r\nhighlighted. Data have been collected during the final experimental\r\nphase of the EU interactIVe project, in which a vehicle\r\ndeveloped by Centro Ricerche Fiat has been used to demonstrate\r\ndriver continuous support produced by an artificial codriver,\r\nwithin a shared initiative framework. 24 subjects drove\r\nthe vehicle on a test route twice: once with the system active,\r\nthe other with the system silent. The test route is composed of\r\nurban arterials, extra urban and motorway roads, and takes\r\napproximately 40ââ?¬â??45 min to be driven. The total database thus\r\namounts to ~35 h of driving data recordings, for a total of\r\n~1.2 M samples per signal. Statistical summary data are\r\npresented, which describe human preferred accelerations, correlation\r\nbetween acceleration, curvature, and speed, and between\r\nlongitudinal and lateral acceleration. Different driving\r\nmodalities, corresponding to different motor strategies and\r\nprimitives, are revealed. Comparisons with literature data are\r\nalso made and discussed. The summary statistics may be\r\nuseful for the design of future ADAS systems, and indeed\r\nthey have been collected for the final tuning of the interactIVe\r\nco-driver.
Loading....