Drivable road detection is crucial in planning routes and prevention of road accidents for autonomous vehicles. The system classifies images captured by camera into two possible scenarios. i) Drivable Road, ii) Non-drivable road. The proposed system focuses on detection of drivable roads in different traffic conditions. Fusion of two feature descriptors BRISK and SIFT are used in the system. BRISK and SIFT are rotation and scale invariant. SVM-RBF has provided highest accuracy score of 70.9 %. Computationally effective methods are implemented for classification and detection of drivable road. Accuracy of the system decreases in the scenes where lighting conditions are poor, and shadows are involved. Graphical user interface with sensors and hardware can improve the overall experience.