Research on the Next-generation Autonomous Vehicle, the Core Means of Transportation in the Future

"Recently, many leading automobile manufacturers in the world have competitively announced their plans to commercialize autonomous vehicles by 2020. It will be possible to see autonomous cars drive by themselves while we tend to our routines during everyday commutes. KAIST C-FRIEND Field-Robot Center has actively carried out research on the core technologies for autonomous driving in uncertain environments. The center has also been developing unmanned airplanes, surface boats, and submersibles with a high level of autonomy, cooperation, and survivability. It is expected that such unmanned systems could change our lives fundamentally in the near future."

Development of Core Technologies for Safe and Reliable Autonomous Vehicles

According to a report published by IEEE (Institute of Electrical and Electronic Engineering) in September 2012, about 75% of all the vehicles in the world will be unmanned by 2040. Also, according to Navigant Research in 2013, the number of autonomous vehicles will increase explosively after 2020, with annual sales volume reaching 95.4 million units per year in 2035. Therefore, all major car manufactures including domestic companies thrive to developing autonomous vehicles.

The research team led by Professor David Hyunchul Shim at KIDCS has been developing autonomous ground vehicles since 2009. He realized the great potential of autonomous ground vehicles at DARPA Grand Challenge, a competition in which unmanned cars drive more than 150 miles of desert road. Since 2005, the contest has been sponsored by the Defense Advanced Research Projects Agency, the top military research funding agency of the US Department of Defense.

Professor Shim also thought that the general technologies related to unmanned vehicles would be directly applicable for the development of planetary rovers for the Korean space programs. Therefore, based on his experience of developing an entire UAV system and also his expertise obtained at Hyundai Motor Company, he formed a small team to develop autonomous vehicle technologies only to discover limitations due to funding and resources.
The big break came in 2009, when Hyundai Motor Company announced its plan to hold a competition for autonomous ground vehicles. He organized a team and started to focus on the research related to unmanned autonomous vehicles. His research team consisted of student and faculty members with automobile engineering, electronics, and computer science backgrounds because an autonomous car can be built only by combining cutting-edge technologies of all of these fields.

There were four major research areas: recognition of the surrounding environment using cameras and LIDAR, computing absolute and relative/absolute positions with respect to the surrounding environment, path planning for obstacle avoidance, and velocity and tracking control for generated paths. In particular, he focused on the development of highly reliable algorithms to enable autonomous navigation in outdoor environments where lighting, weather, visibility, and GPS reception conditions are all highly unpredictable.

In December 2009, the research team was given a KIA Soul, which was heavily modified for basic autonomous driving, for six months. The first competition in 2010 demanded an entry level of autonomy to avoid stationary obstacles. For this, the vehicle was capable of speed and steering control, which was integrated with a higher level algorithm for avoiding obstacles detected by laser scanners. The vehicle was also capable of detecting and following the lanes detected by onboard cameras.

In the second competition in 2012, resolving the problems found in the first competition, the onboard computers were completely replaced with newer and more powerful models, and the overall autonomy was greatly improved by overhauling all hardware and software configurations including laser scanners, cameras, and obstacle avoidance algorithms. Therefore the vehicle is now capable of not only driving along the designated road with stationary obstacles but also avoiding complex, dynamic obstacles, visually detecting signal lights and pedestrians, and parking by itself.

His research team is currently working very hard developing a new autonomous vehicle for the third Hyundai competition in 2014. This time, the team is developing an autonomous vehicle based on the Hyundai Veloster, a sports car, with better acceleration/deceleration and handling. This vehicle is equipped with a higher level of autonomy and reliability for autonomous driving on real roads at higher speed. In the near future, the technologies implemented on this car can be used for an autonomous taxi that can take passengers to the destination by automatically generating a path from the starting point. Since autonomous vehicles, regardless of the based platform, share many component technologies especially at the higher level, Prof. Shim’s research group is simultaneously working on autonomous technologies on unmanned aerial vehicles and autonomous cars, which can be operated together for complex missions.


Prof. Shim, Hyunchul
2013 Annual Report