IMPLEMENTATION OF LONGITUDINAL AND LATERAL CONTROL ON AUTONOMOUS VEHICLE

• Main Author: Adamsyah Ma'ani, Fadillah
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/50131
• Institution: Institut Teknologi Bandung

The plan to use autonomous vehicles in Indonesia’s new capital, along with the high number of accidents caused by human-error, accelerate autonomous vehicle researches in Indonesia. However, autonomous vehicle researches in Engineering Physics Institut Teknologi Bandung (TF ITB) are still done within lab-scale due to the scarcity of real scale platforms. This final project attempted to develop actuation systems (low-level controller) along with longitudinal and lateral control systems (high-level controller) for autonomous vehicle real scale platform based on Yamaha YDRE 2011 golf cart as a foundation for autonomous vehicle researches in TF ITB. The actuation system acted as a low-level controller and consisted of traction, braking, and steering control systems. In this final project, mechanical and electrical systems along with control algorithms are designed and implemented to enable drive-by-wire alongside the existing manual driving systems. The longitudinal control system is designed using a Proportional-Integral-Derivative (PID) controller while the lateral control system is designed using Stanley and PID controller, all implemented using Python programming language. Feedforward also added to the lateral control to improve tracking performance on curved paths. Robotic Operating System (ROS) is used as a communication protocol to connect low-level and high-level controller. The feedback system for the high-level controller is built to estimate the position, orientation, and velocity of the vehicle using sensors such as Global Navigation Satellite System (GNSS) and Inertial Measurement Unit (IMU) along with Extended Kalman Filter (EKF) algorithm. Simulation is done in CARLA Simulator using a waypoint with high-curvature and high-speed characteristics. The resulting lateral, yaw, and velocity error are minimum with Root Mean Square Error (RMSE) of [0.464m; 3.696o; 0,323 m/s] for Stanley controller and [0.822m; 3.076o; 0.381 m/s] for PID controller. The longitudinal controller test resulted in a velocity RMSE of 0.260 m/s. Real-life experiments of lateral controllers are done using three waypoints: a straight track, a high-curvature track, and a low-curvature track. With small initial lateral error, both controller can stabilize the position and yaw of the vehicle with RMSE lateral and yaw of [0.079; 0.178; 0.126] m and [3.509; 5.937; 4.213]o for Stanley controller along with [0.151; 0.160; 0.127] m and [5.147; 4.015; 3.466]o for PID controller. Experiments of both controllers with a large initial lateral error are done using the straight and low-curvature track. From those experiments, RMSE lateral and yaw are obtained as [0.799; 1.290] m and [27.445; 19.856]o for Stanley controller along with [1.263; 1.614] m and [25.984; 30.523]o for PID controller. The traction control system is implemented using a Digital-to-Analog Converter (DAC) and able to actualize throttle voltage accurately with a resolution of 0.0012 Volt. The brake control system is implemented using a DC motor and linear transducer and was able to actualize braking with an accuracy of +0.025 cm and maximum braking time of 0.487 seconds. The steering control system is implemented using a stepper motor and absolute encoder and was able to actualize steering angle with an accuracy of +0.3o and maximum steering time of 7.75 seconds within the steering range of [-35o, 28o]. Finally, this final project has attempted to build low-level controller systems of traction, brake, and steering systems to enable a drive-by-wire scheme and high-level controller systems to enable lateral and longitudinal autonomous path following scheme in a real-scale autonomous vehicle. Performance indices of the controllers used in this final project are presented to be used as a baseline for future implementation and optimization. The usage of ROS, Python programming language, and online repository are hoped to establish a good basis for sustainable autonomous vehicle researches in TF ITB.