METHOD AND SYSTEM FOR LANE TRACKING FOR AN AUTONOMOUS VEHICLE

Disclosed subject matter relates to field of autonomous vehicles that discloses method and system for training a lane tracking system and lane tracking for an autonomous vehicle. During training phase, lane tracking system receives ground truth and measured values corresponding to lane boundary detection points and determines ground truth and measured clothoid points. Thereafter, coefficient values of clothoid parameters are determined for measured clothoid points to model lane boundaries and then Kalman filter parameters are determined for coefficient values to track lane boundaries, using at least one neural network. Further, coefficient values of clothoid parameters are updated using Kalman filter parameters, using which, measured clothoid point is reconstructed, and used to track lane boundaries. Further, training error is obtained by L2 norm between clothoid points and not by coefficients, thereby enhancing training error minimization. The trained lane tracking system is then deployed for lane tracking in the dynamic environment.