Cognitive load estimation using ocular parameters in automotive

•We proposed a system for estimating cognitive load of drivers due to performing secondary tasks while driving.•We proposed metrics to estimate cognitive load from eye gaze and pupil dilation parameters.•We evaluated our pupil-based metrics in estimating cognitive load under varying ambient light conditions.•We evaluated efficacy of our metrics in both simulation and real driving environment involving professional drivers.•Our neural network model classified driving task and secondary task with an accuracy of 75%.