|
Autonomous Driving based on Modified SAC Algorithm through Imitation Learning Pretraining |
|
Mengyi Gao, Dong Eui Chang(KAIST, Korea) |
| In this paper, we implement a modified SAC algorithm for autonomous driving tasks using the simulator AirSim’s environment API which provides various weather, collision, and lighting choices. Given current image state and car velocity as our inputs, the task outputs the throttle, brake, and steering angle data and gives the vehicle action instruction through the AirSim control outputs. As autonomous vehicles are more likely to be accepted if they drive like how human would, we at first train our model by imitation learning to provides the pre-trained human-like policy and weights to SAC. |
|
|
Natural Speed Planning and Control of the Autonomous Vehicle within Social Norm |
|
Solyeon Kwon, Kyoungseok Han(Kyungpook National University, Korea) |
| The tradeoff between strict regulation of autonomous vehicle-driving rules and natural behavior is sufficient to
realize autonomous driving technology. However, most autonomous vehicles’ control logic is designed to strictly satisfy the driving rules that can sometimes cause car accidents and unnatural behavior. In this study, by relaxing the constraints in the optimization problem, we propose the controller, which makes the autonomous vehicle natural behavior, even though it slightly violates the social norm. |
|
|
Vehicle Localization Using Radar Calibration with Disconnected GPS |
|
Jeong Sik Kim, Woo Young Choi, Yong Woo Jeong, Chung Choo Chung(Hanyang University, Korea) |
| As autonomous driving technology develops, research on localization methods is becoming more important. In this paper, we propose global positioning system (GPS) and radar calibration method, and vehicle localization method using a radar sensor based on vehicle to everything (V2X). For vehicle localization, we first propose GPS and radar calibration, which is a way to solve the differences between detection points. With this calibration, during disconnection of GPS, we calculate the position of the ego vehicle by using the rotational transform with vehicle chassis data and radar data. |
|
|
Robust Model Predictive Control with Control Barrier Function for Nonholonomic Robots with Obstacle Avoidance |
|
Ying Shuai Quan, Jin Sung Kim, Chung Choo Chung(Hanyang University, Korea) |
| In this paper, we propose a Robust Model Predictive Control combined with Control Barrier Function (RMPC-CBF) for a nonholonomic robot with obstacle avoidance subject to additive input disturbances. Both guarantees for Input-to-State Stability (ISS) and Input-to-State Safety (ISSf) are provided. ISSf-CBF based safety conditions are formulated as constraints inside a robust MPC strategy. Robust satisfaction of the constraints is ensured by tightening the state constraint set. ISS and robust recursive feasibility are guaranteed by computing the terminal region and state constraint set. Numerical simulation results confirm the effectiveness and validity of the proposed approach. |
|
|
Considerations for Cyber Security Implementation in Autonomous Vehicle Systems |
|
Kyungsu Lee(Infineon Technologies, Korea) |
| Security threats in autonomous vehicle systems are increasing day by day. Unlike systems in other industries, the damage caused by increased security threats in autonomous vehicle systems is directly related to the driver and pedestrian's life. The implementation of cyber security in autonomous vehicle systems is an essential element. Representatively, UNECE adopts regulation on CSMS , making it mandatory to implement cyber security for vehicles and apply the management system. The paper provides considerations for cyber security implementation based on UN Regulation No. 155 such as secure boot, secure communication and secure debug in autonomous vehicle systems |
|
|
A Pallet Recognition and Rotation Algorithm for Autonomous Logistics Vehicle System of a Distribution Center |
|
Kyeongjin Joo, JEONGWON PYO, ARPAN GHOSH, GUNGYO IN(Sungkyunkwan University, Korea), Tae-Yong Kuc(Sung Kyun Kwan University, Korea) |
| This paper introduces a pallet recognition and rotation measurement algorithm for logistic AGV, Based on YOLO v3 and depth camera. From the camera 2D data is collected and the coordinate system from the image is changed into the corresponding world coordination system with the aid of depth data from RealSense camera. Furthermore, by processing point cloud and RANSAC algorithm, we can get the precise orientation and direction of the pallet. Finally, in the experimental result, the autonomous driving AGV with the proposed algorithm shows the results of path planning, loading and unloading the pallet.It also shows that the developed algorithm is applicable to autonomous driving of AGV. |
|