Cognitive Characterization of Automatic Takeover of Driving on Grassland Roads Based on ERP
Exploring the cognitive characteristics of drivers during L3-level conditional autonomous driving takeovers on grassland roads is important for enhancing the universality and safety of autonomous driving technology.Utilizing a load level evaluation meth-od based on sensory engagement dimensions,four kinds of non-driving related tasks(NDRTs)with varying load levels,i.e.,level 0 load(no NDRTs),level 1 load(listening to news),level 2 load(watching videos),and level 3 load(playing games),were de-signed.Simulated driving experiments were also conducted in conjunction with a typical risk takeover scenario(livestock occupying the road)on a grassland road.Event-related potential(ERP)was used to assess the changes in brain topography,latency and wave amplitude values of P300 and N2b in drivers under different NDRTs.Additionally,the comprehensive NASA-TLX scale was em-ployed for both subjective and objective evaluations of takeover cognition.The results showed that,the changes of P300 and N2b components in the frontal and parietal regions,which are involved in cognitive resource allocation and attention regulation,were most pronounced.Both the P300 wave amplitude and N2b latency reached their minimum values under level 1 loading NDRTs,while level 3 loading NDRTs resulted in a significant decrease in P300 latency and N2b wave amplitude.During the grassland high-way autopilot,it was recommended that the driver engaged in level 1 loading NDRTs that utilized auditory and cognitive resources,while avoiding level 3 loading NDRTs that demanded multiple sensory engagement dimensions.
Grassland highwayAutomatic drivingTakeover cognitionNon-driving related tasks(NDRTs)Event-related potential(ERP)
NETL
NSTL
万方数据
