Autonomous Vehicle Research
Mapping and Localization
Our lab studies LiDAR-based 6D SLAM technology for accurate and efficient mapping of an autonomous vehicle. Especially, we approach to solve the various problems encountered when mapping, especially large areas, which are difficult to solve with the conventional SLAM algorithms. We have developed the technology of the Human-in-the-Loop Loop Closing system in which users steps in to optimize the map to build a map efficiently and precisely.
And also we consider highly precise localization algorithm for an autonomous vehicle in urban areas. We study filtering-based localization algorithms using map, such as Particle Filter and Kalman Filter, by fusion of various sensors (GPS, IMU, LiDAR).
As one of the core technologies of autonomous vehicles, path planning provides a trajectory of an autonomous vehicle to follow, which allows an autonomous vehicle to reach the destination without collision. On the road, autonomous vehicles should follow the driving rules and move safely together with other vehicles, avoiding collision with pedestrians and obstacles that can appear suddenly. Unfortunately there is no path planning algorithm that can deal with every possible situation alone. Therefore, multiple path planning algorithms should cooperate closely for safe and stable autonomous driving. In IRiS lab, we are working on fusing various path planning algorithms in order to deal with a wide variety of autonomous driving situations in urban environments.
Vehicle Platform and Control
IRiS Lab has been developing a commercial vehicle-based autonomous driving platform in order to participate in autonomous vehicle competitions. Since 2009, we have retrofitted several commercial vehicles into robotic car platforms, and we named them PHAROS. Those platforms encompass robotized steering, acceleration and braking interface, various sensors, such as lidars, cameras, and GPS, a computing system to analyze the sensory data and perform autonomous driving SW modules, an additional power generation system for the sensors and the computing system, and a networking system based on CAN. Those platforms are currently used for the analysis and validation of the developed mapping, localization and planning algorithms. In addition, we are trying to utilize our autonomous driving platform to perform various research for smart mobility such as, but not limited to, teleoperated driving, self-driving shuttle.