News Flash!!
- We won 2nd place at MBZIRC 2024 with Prof. Jinwhan Kim at ME!! The Prize is 0.5Million USD
- Prof. Shim appeared on EBS and YTN Interview Programs
- We won 2nd place at Hyundai Autonomous Challenge 2023!!
- We won Technology Innovation Award from ROK Army Dronebot Challenge!!
- Our joint research on humanoid pilot robot (PIBOT) is covered by national major news media
- Our journey continues at the historic Monza track!! Watch our Indy Car zipping through the track!!
- We won the 1st place of 2021 Hyundai Motor Challenge!!. 2021년 현대자율주행챌린지에서 우승했습니다!!
우리 무인시스템 연구실은 딥러닝, 항법, 제어기술을 융합한 드론, 자율주행차, 지상로봇과 같은 지능형 로봇을 연구합니다.
우리 연구실은 2020년 과기부 인공지능 챌린지에서 우승한 드론기술을 발전시켜
in-house 3D LiDAR SLAM을 개발해서 미국 NASA JPL 연구진과 협력하여 2021년 9월 DARPA Subterranean Challenge에도 참여하고 자율탐사 지상로봇 개발에도 성공하였습니다.
2021년 미국 Indy Autonomous Challenge에도 현대자동차와의 스폰서쉽 체결결을 통해 자율주행기술의 고도화에도 노력하고 있습니다.
우리 연구실은 real world에서 실제로 동작하는 로봇을 개발합니다. Real World Engineering에 관심있는 학생들을 환영합니다.
Welcome to our homepage. We are performing active research and development of highly advanced
autonomous vehicles by combining various principles of control theory, robotics, and artificial intelligence.
Our effort hopefully will lead to the development of unprecedented groups of intelligent vehicles in near future.
카이스트 전기및전자공학부 정교수
AI 대학원(-22.8), (현)로봇학제, (현)미래자동차 학제 겸임교수
KI 로보틱스 연구소장('19.8~'22.8)
KAIST 미래모빌리티 중점연구소장('21~'22)
국토부지원 민간무인기안전운항연구단장('16-'21)
지능형 무인기 특화연구실장('16-'20)
Professor, School of Electrical Engineering, KAIST
AI Graduate School(ended), Robotics, Future Vehicle Program Professor
Former Director, KI Robotics Institute, KAIST Future Mobility Research Institute
Former Director, Korea Civil RPAS Research Center supportef by MOLIT, Korean Government
Former Director, Intelligent UAV Research Laboratry
< Development of a Mobile Robot System for Delivery in an Indoor Multi-Floor Environment >
< Segmented Map-Based Topological Exploration with Adaptive LiDAR-Inertial Odometry >
< Scaleable and Learnable Autonomous Driving System: Design and Field Evaluation >
< Sensor prediction of missile flight tests >
< Towards fully autonomous racing : system design for the indy autonomous challenge and imitation learning approach using hierarchical policy abstractions >
< Development of an anti-drone system using a deep reinforcement learning algorithm >
< A Deep Reinforcement Learning based Motion Planner for the Exploration of Unknown Environments with an Aerial Robot >
< Detect-and-Avoid Technology Development and Flight Test for Unmanned Aircraft System Integration >
< (A) Scalable Path Planning Algorithm for Drone Swarms using Density Control >
< Design of an Aerial Combat Algorithm Based on Basic Fighter Maneuvers and Reinforcement Learning >
< Research on the actuator fault tolerant adaptive control scheme to enhance safety of multicopter>
< Research on the Image-based Visual Servoing Framework Considering the Aerial Manipulation >
<Development of urban self-driving vehicle using lane detection on the road and localization in intersection with lane map>
-
< A Multimodal Flight Control Design and Flight Test of a Tail-sitter UAV >
< Design of an Aerial Combat Guidance Law using Virtual Pursuit Point Concept >
< Vision and LIDAR-based Autonomous Navigation System for Indoor and Outdoor Flight of Unmanned Aerial Vehicles >
< Research on the UAV Model Based Design and Adaptive Control for Uncertain Dynamics >
< Research on the UAV Model Based Design and Adaptive Control for Uncertain Dynamics >
- 모바일 매니퓰레이터를 활용한 텔레프레즌스 구현
- 모바일 매니퓰레이터의 실내환경 충돌 회피 경로 생성 및 제어
- 화학 작용제가 존재하는 실내환경에서 드론과 로봇을 활용하여 정찰 및 오염지도 생성
- 드론-로봇 협업 시스템을 활용한 오염된 실내환경 자율 제독 시스템 구축
- 고성능 전기차의 서킷 주행 성능 향상을 위한 공력 제어 시스템 개발
- 차량 모델 기반의 4-코너 독립 공력 제어 알고리즘 개발 및 레이싱 트랙 실증
- MPC를 활용한 공력 제어 시스템의 응답 지연 보상
- 실내 복잡 환경(공장 환경 등)에 적용가능한 실내 정밀 측위 및 자율주행 기술 개발
- 자동화 시스템 운용을 위한 하드웨어, 소프트웨어 아키텍처 개발
- 모방학습 및 강화학습 기법을 복합적으로 활용하여 Sim-to-real gap이 적은 정책 모델 학습 기법 개발
- Transformer 구조 기반 자율 주행 정책 모델 설계
- 실 환경에서의 실험을 통한 학습 모델의 Sim-to-real 성능 검증
- LiDAR 기반 3차원 위치 추정 알고리즘 개발
- 동적 및 정적 장애물 회피 알고리즘 개발
- 효율적인 탐사 알고리즘 개발
- 드론 연계 실내외 도심환경 자율주행 배송 기술 개발
- 다중 로봇 동시 운용 기술 고도화
- 불법드론에 대한 지능형 무력화가 가능한 신속대응형 드론 개발
- 초도 탐지 기술 및 불법드론에 대한 상시순찰형 드론 개발
- Robot Operating System을 이용한 융합센서 운용/임무제어 시스템 개발
- 대상 수신기의 GPS 신호 끊김없이 스푸핑이 가능한 GPS soft 스푸핑 신호 생성 SW
- 드론의 회전, 직진 및 선회 기동성을 활용한 방탐 및 위치 추적
- 융합센서/재머 탑재한 드론 구성
- 다중 에이전트를 위한 표준 인터페이스 학습 프레임워크 구축
- 다중 에이전트 강화학습 모델 프로토타입 개발
- 검증용 시나리오를 통한 다중 에이전트 모델 효용성 확인
- 항공기 운항에 전단계{Taxing, Takeoff, Cruise, Landing, Emergency situation}에 대응 가능한 인간형 로봇 조종사 개발
- 자연어 조종 매뉴얼을 학습하여 지식베이스화, 이를 바탕으로 추론, 상황에 적합한 행동을 직접 판단하는 Intelligent Robot System 구현
- 실내 환경에서 주변 환경 인식률을 향상하기 위한 센서 융합 기반의 인식 알고리즘 활용
- 다층 환경에서의 경로 계획을 위한 통합형 정밀 지도 구축 및 항법 수행
- 팔레타이징 로봇 연계 인터페이스 및 도킹 알고리즘
- 자율주행 로봇에 탑재된 매니퓰레이터를 이용해 실내환경에서 문, 엘리베이터 같은 시설물을 스스로 통과하는 시스템 개발
- 비정형 실내 환경에서 자율주행 로봇의 자율성 고도화
도심 환경에서 배달을 수행할 수 있는 자율주행 로봇을 만드는 프로젝트입니다. 복잡한 환경에서도 안정적으로 주행하기 위해 지형 정보를 이용하는 node-link 개념을 활용하여 로봇이 목적지까지 안전하고 빠르게 도달할 수 있습니다. 주변 환경을 인식하는 다양한 센서의 조합을 통해 도심 환경에서도 강건한 항법 알고리즘을 개발하고 있습니다.
Hyperparameters significantly affect the intelligent robotic system performance. However, the explicit relationship between them is not known (Blackbox). The system evaluation can be done only by sampling, which is usually expensive. The objective is to find the systematic way to ease the system optimization.
라이다를 이용한 Detection, Tracking and Prediction
- 딥러닝을 이용한 물체 인식, 추종(tracking-by-detection)
- Convolutional GRU를 이용해 타 차량의 궤적 예측
생산 현장에서 적용 가능한 자율주행 로봇의 알고리즘 개발 및 적용
도로 환경의 주변 전력장치 검진 시스템 개발
- 차량 시스템을 위한 전력 시스템 설계
- 센서 시스템 개발
- 시스템 Monitoring을 위한 HMI 개발
전자과 내 인공지능 연구실과 함께 클러스터로 구성되어 자율주행 구현을 담당, 도심 자율 주행을 위한 인공지능 기술 기반 알고리즘 개발 및 자율 주행 시스템 개발
While there are impending explosive demands, in order to safely integrate unmanned aerial vehicles into
civil airspace, - Prepare critical infrastructure for basic operation of civil RPAS - Collect flight data
of RPAS under various experiment conditions using the aforementioned infrastructure, - Analyze the
collected data to draft flight safety regulations, certification process, and operation procedures By
executing the procedures listed above, our nation can be prepared for the impending era of civil RPAS and
participate in the ICAO RPAS rule making process as a leading country based on the research results
generated from the domestic RPAS operation.
This project is funding by Ministry of Land,
Infrastructure, and Transport from 2016.
The objective of this project is establishment of the USV(Unmanned Surface Vehicle) certification system
and regulation for the operation of USVs in Korean waters and development of cooperative technology for
multiple heterogeneous unmanned vehicles(USV and UAV) and technology for their maritime application
services
This project is funded by the Ministry of Oceans and Fisheries from 2015.
The objective of this project is that we develop the smart drone. Smart drone can be used in a variety of
fields related with safety and convenient service by using ICT technology. Especially, Our research lab
develop the vision-based automatic landing, fault tolerant controller algorithms for the rotary
UAV.
This project is funding by Ministry of Trade, Industry and Energy from 2016.
- Vision-based aircraft detection using deep learning
This project aims to develop airborne collision
avoidance system and related technology based on the performance of manned aircraft for civil aviation
airspace integration operation. In this study, a remote aircraft is detected using an image sensor. Deep
learning technology enables detection of flight objects in cluttered backgrounds and has a fast processing
speed that can be used in embedded computers.
This research is funded by Uconsystem and the Ministry of
Trade, Industry & Energy from 2015.
Autonomous driving and artificial intelligence technologies are the leading fields of the 4th industrial
revolution. In this project, our consortium aims to develop the EV based open autonomous vehicle platform
that allows ♂access to the essential technologies for autonomous driving such as perception, planning
and control. Among them, our research team is performing autonomous driving control based on End-to-End
deep learning ♂which predicts optimal control command from sensing data as one neural
network.
This project is funding by Ministry of Trade, Industry and Energy from 2017.
Recently, drone market increased rapidly according to the growing popularity of drones. However, as the usage of drones increased, the people who take advantage of the drone increased as well, from minor violation like picturing the private life with the camera on it, to a special crime like delivering drugs. This research aims to develop the vision-based UAV to track and capture the target, which is the drone attacking to our camp. To achieve the objective, our research team integrate the autonomous flight system and the on-board image-processing system on the one UAV platform.
The purpose of this research is development of a humanoid robot acts as the pilot for converting existing aircraft or car into unmanned aircraft or unmanned car with minimal modifications. In contrast to current trend, such as OPV technology and auto-pilot, this proposed method could give full-authorization to pilot robot and decrease altered parts and the required time. The pilot robot hardware is composed of four 6-DOF manipulators(two for arms and two for legs), two adapters for both hands, body frame, vison sensor. The software architecture is designed to automate flight operation from take-off to landing. The core of software is feedback control of aircraft by manipulating cockpit components based on the flight states received from flight simulator computer and waypoint planning.
The purpose of this research is development of the autonomous robot which mutually supplements both issues of the UGVs and the UAVs by convergence of both systems. Unmanned ground robots have demerits about relatively short mission range and getting over the obstacles, and UAVs have issues of noise problem and lower energy efficiency. The developed robot can achieve various ground/aerial mixed missions which cannot be achieved by existing unmanned robots.
The UHV has ALFUS autonomy level 4, then it decides actions throughout a mission and notices to a user. It will be basically used for the surveillance and reconnaissance, or measurements of the radioactivity, gas, temperature, and humidity with additional sensors. This project demands overall unmanned system techniques such as indoor/outdoor navigation, guidance, obstacle detection and avoidance, environment modeling, path planning, etc.
Recently, there have been significant advances in self-driving cars, which will play a key role in future intelligent transportation system. In order for these cars to be successfully deployed in real roads, they must be able to drive by themselves along collision-free paths while obeying various traffic laws. In contrast to many existing approaches that use pre-built environment maps of roads and traffic signals, we propose system using a unified map contains not only the information on real obstacles nearby but also traffic signs and pedestrians as virtual obstacles. Using this map, the path planner can efficiently find paths free from collisions while obeying traffic laws.
EureCar is a Self-driving car that can drives itself along the pre-planned path while avoiding obstacles and obeying various traffic laws. We have been developing 2 Self-driving cars, EureCar and EureCar Turbo. EureCar (our first Self-driving car) has high precision positioning system, 7 laser scanner and 4 camera has been developing for 2 years. EureCar Turbo takes 6 months of development time for the same performance as EureCar with less and low-cost sensors. We could curtail the development period from 2 years to 6 months by using previous software developed for EureCar. Especially, it took only 2 months for developing additional software for EureCar Turbo except the period of hardware implementation and remodeling.
We are developing a 3D navigation algorithm to provide a MAV with the estimated pose information in real-time. The navigation algorithm is based on Monte Carlo Localization (MCL) using a particle filter. The particle filter calculates likelihood of particles using laser measurements and ray-casted ranges of the particles that uses an octree structure for fast computation. The MCL using a particle filter is implemented to estimate position and attitude of the vehicle.
To validate the proposed indoor navigation algorithm, the experiment is conducted using a quad-rotor platform, a laser scanner, an IMU, a low-level flight controller, and a ground station computer. The onboard flight computer runs a low-level controller for position and attitude control. The attitude and position is controlled by a multi-loop PID controller and the guidance algorithm to follow waypoints or paths is also enabled. In addition, obstacle avoidance is available to deal with sudden interference of human for safety.
Our proud Indy race platform (yes, we own it!!)
on-campus coalition
KAIST insitute for Robotics (KI-R) (~Aug 2022)global collaboration
Team CoStar for DARPA Subterranean Challenge We collaborated with Dr. Ali Agar at NASA JPL for DARPA Subterranean Challenge from 2018 to 2021.3D 거리센서(LiDAR)와 2D 이미자센서(Omni-camera, Ladybug5) 융합 및 인공지능 기법(Semantic segmentation)을 활용한 야지(비정형 지형) 환경 인식 기법 개발
- Sensor fusion을 위한 구조물 설계
- Sensor fusion algorithm 개발 및 적용
- Deep learning 기반의 Semantic segmentation 기법 개발 및 적용
In this research, we aim to develop guidance algorithms that enable highly maneuverable unmanned combat
aerial vehicles(UCAVs) equipped with thrust vectoring to engage a close-range aerial combat. The proposed
three-dimensional guidance algorithm is complemented by a precision trajectory tracking, which is robust
to damages or malfunctions of control surfaces or airframes. The proposed guidance and control algorithms
will be validated in a series of high-fidelity nonlinear simulation using MATLAB and X-Plane and then real
flight experiments with small-size UCAV test platform.
This project is funding by National Research
Foundation (NRF) from 2011.
In this research, The stability and control augmentation system and longitudinal auto-landing guidance
law, using Time Delay Control, is proposed and evaluated through a simulation with model uncertainties and
wind disturbances. And the TDC is validated by a series of flight experiments.
This project is funding
by Korea Aerospace Research Institute(KARI) from 2010.
There have been ongoing interests in a type of aircraft that are capable of vertical take-off/landing (VTOL) for greater operability and high-speed horizontal flight capability for maximal mission range. A possible solution for such application is tail-sitters, which takes off vertically and transitions into a horizontal flight. During the entire mission of a tail-sitter from take-off to landing, it goes through largely varying dynamic characteristics. In this project, we propose a set of controllers for horizontal, vertical, and transition flight regimes. Especially, for transition, in conjunction with conventional multi-loop feedback, we use L1 adaptive control to supplement the linear controllers. The proposed controller were first validated with simulation models and then validated in actual flight tests to successfully demonstrate its capability to control the vehicle over the entire operating range.
Unmanned Aerial Vehicles(UAVs) have been performed important role in modern wars. Several countries including United States are accelerating developments on the unmanned aircraft systems while revealing their experimental works on the unmanned systems. Unmanned agents have been applied in many warfare situations such as reconnaissance, surveillance or retrieving land mines. In the future, however, multi and heterogeneous unmanned platforms will perform various cooperative tasks such as assault missions, which are managed by high-level commands from a commander. We are therefore doing a research on developing such a cooperative system for heterogeneous UAVs. We are focusing on designing a system to manipulate several fixed and rotary wing UAVs simultaneously in order to provide availability of cooperative missions.
The aim of Intelligence-based Unmanned Aerial Vehicle Control Specified Research Center is developing the original technology of UCAV(Unmanned Aerial Combat Vehicle). Specified research center뭩 research objective is establishing the theoretical structures and related technology bases for single/multi UAV뭩 effective fulfillment of surveillance/reconnaissance/information gathering/ attack in future warfare. UAV used for reconnaissance and transportation conducts a key role in current warfare. In the near future, UAV is expected to play a key role in warfare by conducting missions such as destroying enemy뭩 air defense network, winning the air and other important tasks with sufficient maneuverability and attack ability. To reduce RCS(Radar Cross Section), UCAV requires specialized wing and body shape which causes instability of aircraft. Also, aerodynamic analysis of specialized wing and body is difficult. Thus, development of new smart flight control device, task design, decision making and effective ground system must be preceded for UCAV to be controlled stably and to perform attack missions which can be very complicating.
The specified research center뭩 key subjects of research are shape design of UCAV which has minimum RCS and to secure the original technology for smart guide control and decision making algorithm for improvement of performance ability. It is required to research the smart decision making algorithm, shape of UCAV for LPI(Low Probability of Intercept)/high agile maneuver, smart control system based on piloted handling and ground system for management of multiple UAV(Unmanned aerial vehicle).
The researches for UCAV have been conducted narrowly in South Korea. Thus, to secure technologies of related areas, establishment of specified research center for the development of UCAV technology and training of professional manpower was required. To meet these requirements, this specified research center conducts research on 3 detailed assignments. 3 detailed assignments are as follows;
SU-1 : A Study on Smart Decision Making Algorithm for Unmanned Air Vehicle
SU-2 : A Study on Smart Control System for Air-Combat-Maneuvering of Unmanned Aerial Vehicle
SU-3 : A study on the technology of Human-Machine Interaction to reduce load of multiple UAV operator
SU-1 : A Study on Smart Decision Making Algorithm for Unmanned Air Vehicle
The purpose of a Study on Smart Decision Making Algorithm for Unmanned Aerial Vehicle is to develop smart decision making algorithm to improve the mission success rate in response to uncertain and dynamic using environment, risk elements and changing of aircraft뭩 condition. To achieve this purpose, study on study items, as follows, will be conducted.
> Study on decision making algorithm architecture of intelligent-UCAV
> Study on system monitoring algorithm of intelligent-UCAV
> Study on profile scheduling and task scheduling/re-scheduling algorithm
> Study on intelligent learning algorithm
SU-2 : A Study on Smart Control System for Air-Combat-Maneuvering of Unmanned Aerial Vehicle
The purpose of a Study on Smart Control System for Air-Combat-Maneuvering of Unmanned Aerial Vehicle is designing UCAV뭩 shape satisfying LPI/High mobility and evaluating this design and then, using specified configuration, suggesting 6DOF(Degree Of Freedom) guidance model and designing intelligent control system based on piloted handling. To achieve this purpose, study on study items, as follows, will be conducted.
> Aerodynamic analysis for aircraft뭩 controllability and configuration design based on analysis of low RCS.
> Designing aerodynamic database about LPI configuration and constructing 6 DOF model for aerodynamic analysis and handling performance.
With Grand Challenge and Urban Challenge held by DARPA in 2004,2005,and 2007 and new car techonologies being introduced that provide drivers with safe and conveinent driving such as parking assistance, cruise control and lane keeping system, Autonomous Car became one of the most popular research area recently.
EURECAR, based on SOUL from Hyundai & KIA Motor Company, is our first autonomous vehicle developed. It can travel without any human control using GPSs and an INS for localization, vision sensors including cameras and laser scanners for local localization and obstacle detection, path-planning algorithms to generate a path that should be tracked, and actuators and their controllers to track the path with good performance. A new version of EURECAR (isn't named yet) with better performance is being developed based on Veloster Turbo also from Hyundai & KIA.
Future combat system will be carried out by a large number of heterogeneous groups of both manned and unmanned sub-systems, which interact one another in systematic, hierarchical, and highly organic manners, so that a mission can be carried out with minimal collateral damage while maximizing the outcome. Therefore, future warfare will be carried out by groups of hierarchical and organically integrated manned and unmanned groups and therefore research on the integration and operation of heterogeneous multi-agent system must be preceded. Multi-agent scenarios require more sophisticated and flexible operation techniques to adapt to complex and less-predictable situations than conventional single-agent-based approaches. Before each mission, the given mission is analyzed, planned, allocated to the participating agents and then individual path for each agent is planned to the best knowledge available. During the mission, each UAV avoids collision with other agents and other obstacles by dynamically replanning their paths using the map augmented using the sensor measurement. Participating UAVs need to cooperate and synchronize their actions based on the information collected, processed, inferred, and shared. Their missions are then replanned and reassigned in order to adapt to the changes of the surroundings and thus maximize the mission effectiveness.
In this research, we aim to deliver a set of field-deployable research outcomes for the development of distributed multi-UAV system, which is an indispensible component of future network-centric warfare system. We pursue such a goal by researching on the component technologies for efficient operation of multiple UAVs, namely mission planning/assignment, information fusion and inference, dynamic mission allocation, and path planning. The research outcomes on these component technologies will be then integrated, implemented, and validated in a series of simulations and experiments.
- Micro Fin System
Flow on the airfoil section in a wing is separated earlier as the angle of attack is
increased. And the wake after separation makes the drag force bigger so that the more energy is needed to
be used for the flight at high angle of attack. This project researches on the reduction of the drag force
using vortex generator like Micro Fin to delay the separation even at high angle of attack.
- Micro Flap System
A Blended Wing Body(so called, BWB, Flying Wing or Zagi) aircraft doesn’t have any lateral directional
control surfaces like vertical fin, and can’t assure that directional stability so that has adverse yaw
and large dutch roll dynamic characteristics. This point makes the aircraft have little controllability in
that direction. This low stability & controllability in the direction is critical to UCAV(Unmanned Combat
Aerial Vehicle) of that type and of course, should be ensured. This project studies on the augmentation of
controllability of UCAV(or normal UAV) of that type using the Micro Flap which is a control surface for
the yaw direction.
We are developing a mining robot works in underground coal mine has extremely harsh environment to human workers. The robot has to maintain its mechanical reliability under high temperature, high humidity, and lots of dust in the air. Also, the robot can measure the environment map and localize itself by fusing heterogeneous sensors such as stereo camera, laser scanner, and inertial measurement unit. Measured data analyzed intellectually is transmitted to human in the remote control room, so that the human worker could control the robot easily using virtual reality technology and force feedback joysticks. Therefore, development of this mining robot will surely contribute to make a safe workspace for human and to decrease the ratio of industrial accident. In addition, it can increase the productivity of underground resources and resolve the suffering from shortage of labor caused by avoiding 3D works.
Future combat system will be carried out by a large number of heterogeneous groups of both manned and unmanned sub-systems, which interact one another in systematic, hierarchical, and highly organic manners, so that a mission can be carried out with minimal collateral damage while maximizing the outcome. Therefore, future warfare will be carried out by groups of hierarchical and organically integrated manned and unmanned groups and therefore research on the integration and operation of heterogeneous multi-agent system must be preceded. Multi-agent scenarios require more sophisticated and flexible operation techniques to adapt to complex and less-predictable situations than conventional single-agent-based approaches. Before each mission, the given mission is analyzed, planned, allocated to the participating agents and then individual path for each agent is planned to the best knowledge available. During the mission, each UAV avoids collision with other agents and other obstacles by dynamically replanning their paths using the map augmented using the sensor measurement. Participating UAVs need to cooperate and synchronize their actions based on the information collected, processed, inferred, and shared. Their missions are then replanned and reassigned in order to adapt to the changes of the surroundings and thus maximize the mission effectiveness.
In this research, we aim to deliver a set of field-deployable research outcomes for the development of distributed multi-UAV system, which is an indispensible component of future network-centric warfare system. We pursue such a goal by researching on the component technologies for efficient operation of multiple UAVs, namely mission planning/assignment, information fusion and inference, dynamic mission allocation, and path planning. The research outcomes on these component technologies will be then integrated, implemented, and validated in a series of simulations and experiments.
Aircraft are inherently flexible and flexibility can be used to advantage as a design feature for improved performance. Flexible aircraft are subject to interaction between flight mechanics, structural dynamics, and flight control system dynamics. Aeroservoelastic instability can lead to disastrous failures in aircraft as was recently the case for several aircraft. These problems are very similar to flutter accidents which occurred two decades age.
Aeroservoelasticity(ASE) is a multidisciplinary technology dealing with the interaction of the aircraft’s flexile structure, the steady and unsteady aerodynamic forces resulting from the aircraft motion, and the flight control systems. Detailed and quite complex mathematical models are required to accurately predict" ASE interactions" and to design active control systems for flexible vehicle applications.
A compatible mathematical model of the coupled dynamics of flight mechanics, flight control, and structural dynamics must be established. Such models(using the time and frequency domain descriptions) are available in the aeroelastic analysis for aircraft with high aspect ratio flexible wing to assist in the design and clearance of flight control systems. Therefore, the aeroelastic analysis technique for aircraft with high aspect ratio flexible wing, and its use in aircraft design applications are studying.
To employ unmanned aerial vehicles into same airspace with manned aircraft in near future, collision avoidance algorithm for UAV with safety certification will be needed. When other cooperative avoidance systems fail or are not supported, non-cooperative avoidance systems, called sense-and-avoid algorithm such as using radar and camera should be used for detecting and tracking opponent aircraft approaching into UAV's protected volume.
In this research, we develop algorithms and system for collision avoidance of aircraft via vision sensor technology. Various image processing methods and Support Vector Machine (SVM), Kalman filter are used for target detecting and tracking. This Information of target is used for collision avoidance algorithm such as Model Predictive Control (MPC) or potential function approach. Finally, the validation experiments of algorithm for UAV system will be conducted.
The objective of this project is to develop tail-sitter UAV which is complement defects of the existing reconnaissance UAV. The fixed-wing UAV can’t acquire clear imagery intelligence and the rotorcraft can’t approach operation area quickly due to slow cruise speed. Tail-sitter UAVs have vertical capability of a helicopter and the speed comparable to a fixed-wing airplane.
During the take-off phase, the aircraft climb up to adequate altitude vertically. Then, it goes from vertical to horizontal flight. This process is called the transition flight. After transition, the aircraft flies horizontally just like a fixed wing airplane. T he vehicle regains a vertical attitude via the transition flight and then descends slowly to land. Therefore, we research configuration design and multi-mode controller which consist of hover, transition, and cruise to develop tail-sitter UAV.
In this research, we aim to develop guidance algorithms that enable highly maneuverable unmanned combat
aerial vehicles(UCAVs) equipped with thrust vectoring to engage a close-range aerial combat. The proposed
three-dimensional guidance algorithm is complemented by a precision trajectory tracking, which is robust
to damages or malfunctions of control surfaces or airframes. The proposed guidance and control algorithms
will be validated in a series of high-fidelity nonlinear simulation using MATLAB and X-Plane and then real
flight experiments with small-size UCAV test platform.
This project is funding by National Research
Foundation (NRF) from 2011.
In this research, The stability and control augmentation system and longitudinal auto-landing guidance
law, using Time Delay Control, is proposed and evaluated through a simulation with model uncertainties and
wind disturbances. And the TDC is validated by a series of flight experiments.
This project is funding
by Korea Aerospace Research Institute(KARI) from 2010.
The objective of this project which is funded by ADD (Agency for Defense Development) is to design and develop robust and precision guidance and control law for ATOL (Automatic Take-Off and Landing) of MUAV(Medium-altitude long endurance Unmanned Aerial Vehicle) against external disturbances such as wind shear, turbulence and gust. Due to its high aspect-ratio and long wing span, the ATOL system has to consider its aerodynamic and structural characteristics such as Aeroservoelasticity (ASE) and limits of maneuverability in roll during the landing and take-off.
Unmanned technology for almost all forms of transportation (vehicle) such as aerial, ground, and under water vehicles was applied and used for a variety of missions. In the near future, the intelligent unmanned technology is expected to be developed for the unique mission to perform alone, or coordination with the other systems.
Unmanned autonomous ground vehicles create and follow the path for the given waypoints to reach the target area by considering unpredictable internal and external problems. It’s difficult to predict Kinematic / Dynamic constraints unless the vehicle drives mileage and a spacious flat with no obstacles. Therefore, the autonomous systems are required collect information through a number and various kinds of sensors (sensor fusion), reconstruct the environment, generate the safe path and track the path in a predictable time. Especially high-level reasoning skills are required that consider uncertainties and obey traffic laws while driving in the real world.
1st unmanned self-driving vehicle competition will be held under the auspices of the HYUNDAI mobile. The goal is to develop the self-driving car (figure) obeys fair rules and regulations as well as avoid different variety of the obstacles in the road.
This project is a subscale aircraft, K-UCAV which was funded by Korea Aerospae Indeustries(KAI) to conduct air-to-air and air-to-ground missions, including suppression of enemy air defenses. The plane is 20% scale-down model of the K-UCAV that it has built to validate the aerodynamic aspects of its design.The main wing is the type of low-wing with sweepback angle and taper and tail wing is the V-tail type.
In this project, awe develop cutting-edge sensing and guidance algorithms that allow the host vehicle to
land safely to a shock-absorbing net with a visual marker. Such a feature is a great plus to enable safe
operations during the landing stage, during which a high percentage of accidents occur in reality. I am in
charge of the hardware, airframe, and control of the aircraft.
Since the beginning of the project, we
have been able to come up with a slightly different idea of using an inflated airbag. This airbag not only
helps to reduce the shock during the landing, but also serves as a visual marker that can be detected at a
distance using a relatively simpler and more reliable algorithm. We have been able to validate our
algorithm using visual feedback and successfully landed the vehicle at a very high success rate so far.
This is a project funded by Korean Ministry of Commerce, Industry, and Energy from June 2007. The objective is to develop a helicopter that can fly autonomously to perform crop dusting task with minimum involvement of ground crews, i.e., farmers. Therefore the system has to be reliable, efficient, and easy to use. The helicopter is equipped with inertial navigation system, GPS, laser altimeter, and other sensors to ensure safe operation. The ground system is designed to provide intuitive and informative user interface to the functialities of the designed system. This work is a joint work with Oneseen Skytech.
Main Rotor Diameter | 3135 mm |
Height | 1180 mm |
Width | 770 mm |
Powerplant | watercooled 2cycle engine 24hp |
Operating time | 60 min. |
Specification of X-Copter 1
This research is for the development of unmanned aerial vehicles that are capable of long-endurance and long-distance missions using an advanced power technology currently under development in the Department of Aerospace Engineering at KAIST. Although traditional energy sources based on hydrocarbon have reasonably high energy densities, the overall efficiency is still far below an acceptable level and they are the primary cause of air pollution and greenhouse effects. On the contrary, the fuel cell generates electricity directly from the fuel without lossy combustion process. However, the conventional fuel cell technology typically uses pure hydrogen, which has to be compressed to a very high pressure or cooled to liquid state to meet the desired energy density. Such processes are prohibitively costly and hardly justifiable except for few applications such as spacecraft. Our fuel cell technology uses a liquid solution of hydrogen compound as the source of hydrogen and therefore the achievable energy density is significantly higher and the refueling and storage processes are much more convenient than conventional methods using pure hydrogen. Our proposed technology is therefore ideal for many power-demanding field systems including unmanned vehicles. Our UAV testbed is based on a blended wing-body platform, which has a very high lift-to-drag ratio, allowing a large payload weight and volume at a high efficiency. We are working on a long-rannge flight mission that can showcase the performance of our proposed UAV system... stay tuned!
A test flight video can be found here (53MB).
It is of high importance to guide a payload to a landing site at a greater accuracy. This task requires an active control and precision positioning system to guide the payload to reach the target point so that it can be used to deliver the supplies for soldiers in the battlefield or refugees in isolated areas. We consider a parafoil-based system because its airfoil-shaped canopy can generate a lift and steering capability instead of simply braking the descent as is the case with conventional parachutes. We have investigated various possibilities on the mechanism to effectively control the parafoil. We chose Global Positioning System (GPS) as the sole source of position measurement. We designed an algorithm to control the system only using the GPS data to guide the whole system to the target location. We validated this concept by developing flight control software in C++, integrated with a wireless communication system and a servo driver circuit. During flight tests, we used an electrically powered remote control helicopter to drop the parafoil system. A series of experiment show our parafoil control system could guide the payload to the prescribed location with a satisfactory accuracy.
Related paper:
Yeon-Deuk Jeong, Sang-Woo Moon, and Hyunchul Shim, "Development of a GPS-based
Autonomous Parafoil Control System," KAIST-Kyushu University Joint Workshop, 2007
I have been involved in the Berkeley UAV research group, BEAR, since its beginning in 1996, under the guidance of Professor Shankar Sastry (Dean of Engineering at UCB as now). The followings are the achievements with my direct involvement.
Sponsor: UCAR project, DARPA
Duraion: 2004
The objective of this project is to provide a viable solution for autonomous flight in an environment with urban or natural obstacles. The obstacle avoidance problem is resolved by combining the model predictive control approach with the concept of potential field. The MPC algorithm with an efficient gradient-search based optimizer is capable of generating the collision-free trajectory in real time, unlike existing algorithms not capable of real-time trajectory replanning or handling three-dimensional environment. The adjacent obstacles are sensed with an onboard laser scanner, which provides a local obstacle map along the flight path with sufficient accuracy and reliability. Major accomplishments are 1) the development of real-time solver for the proposed MPC algorithm, 2) the effectiveness of the cost function formulation that accounts for obstacle avoidance and trajectory tracking and 3) obstacle sensing and local map building using onboard laser scanner. The proposed framework was successfully tested on a Berkeley UAV as shown in the figure below in a series of test flight, in which the RUAV flies close to the shortest path to the destination while reaming free from any collision with nearby obstacles.
Related Papers
D. Shim,
H. Chung, H. J. Kim, S. Sastry, "Autonomous Exploration in Unknown Urban Environments for Unmanned
Aerial Vehicles," accepted to AIAA GN&C Conference, San Francisco, August 2005.
D. H. Shim, H. J. Kim, S. Sastry, Decentralized Nonlinear Model Predictive Control of Multiple Flying
Robots in Dynamic Environments In IEEE Conference on Decision and Control, December 2002.
I have been involved in the Berkeley UAV research group, BEAR, since its beginning in 1996, under the guidance of Professor Shankar Sastry (Dean of Engineering at UCB as now). The followings are the achievements with my direct involvement.
Sponsor: PAM project, DARPA
Duration: 2003-2004
In this seedling project, BEAR team undertook a
challenge to provide a UAV platform capable of performing a fully autonomous mission from take-off to land
without any human assist. This goal calls for a vehicle with a powerplant that does not need any external
help to start and stop. Conventional radio-controlled helicopters are based on four-stroke gasoline engine
or glowplug engine, which all need external assist to start. Therefore, BEAR team chose an electric
helicopter as the platform, which is powered by extremely large capacity lithium polymer batteries. (40V
8000mAh in total). The brushless DC motor with RPM governing is the perfect answer for our applications-
fully remote start/stop operation is simply done by a flick on the transmitter. The vehicle performs, as
shown in the video link below, fully autonomous take-off and landing repeatedly without any human aid that
we need for gas-powered helicopters.
The following link is a video of Ursa Electra 1, the world's first electric UAV with fully autonomous take-off to land capability: Click here
** this is an unfortunate research activity where my original input about using MPC for aerial pursuit-evasion game went uncredited...
Sponsor: Software Enabled Control (SEC), DARPA
Duration: 2004
This demonstration involved the
development of a model predictive controller that tracked a final waypoint, as well as avoided a moving
pursuit aircraft through the same controller. The timescale of this project was about 6 months. We
successfully demonstrated the algorithms and code against a trained pilot, who was flying an F-15 (see
videos below).
Related Papers:
J. Mikael Eklund, Jonathan Sprinkle, S. Shankar Sastry, "Implementing and Testing a
Nonlinear Model Predictive Tracking Controller for Aerial Pursuit Evasion Games on a Fixed Wing Aircraft",
Proceedings of American Control Conference (ACC) 2005, (In Publication), Portland, OR, Jun., 8--10, 2005.
Jonathan Sprinkle, J. Mikael
Eklund, H. Jin Kim, S. Shankar Sastry, "Encoding Aerial Pursuit/Evasion Games with Fixed Wing Aircraft
into a Nonlinear Model Predictive Tracking Controller", IEEE Conference on Decision and Control,
Submitted, Dec., 2004.
D. H. Shim, H. J. Kim, S.
Sastry, Decentralized Nonlinear Model Predictive Control of Multiple Flying Robots in Dynamic
Environments,?IEEE Conference on Decision and Control, December 2003.
Acknowledgment. This section is adapted from the original page by Jonathan Sprinkle. The original is found here
In this research, the dynamic trajectory replanning in the event of possible mid-air collision is demonstrated. Each vehicle's flight control system is supervised with a online, fully decentralized trajectory generation based on model predictive control. Each vehicle boardcasts their current location. The proposed approach successfully finds safe paths for five helicopters in the simulation shown below. Encouraged by the simulation's outcome, we went ahead to try the idea on real Berkeley UAV platforms, each of which costs $100,000 (material only)!! So in total of $200,000 was set on a collision course. One can say we put two high-end Porches on a collision course and we hope our system can find safe paths to get these vehicles around each other without running into them. The two UAVs, Ursa Magna 1 and Magna 2, are set in an intentional head-on collision course. The trajectory planner based on model predictive control computes the trajectories for two UAVs, which minimize the given cost functions that penalize the proximity to other UAV and the deviation from the original course at the same time.
Upon the successful test, our daring experiment grabbed a lot of attention and Discovery Science channel was one of them. The film crew came out in May 2003 and flimed out experiment, as linked below.
Related Papers:
D. H. Shim, H. J. Kim, S. Sastry, 밆ecentralized Nonlinear Model Predictive Control of
Multiple Flying Robots in Dynamic Environments,?IEEE Conference on Decision and Control, December 2003.
This research was covered by Discovery Science Channel Canada in 2003, as can be seen in the follwing link:
Sponsor: Dr. Alan Moshfegh, Office of Navy Research (ONR)
This project intends to deliver a technology
that enables a rotorcraft-based UAV to land on a ship using its onboard vision system. It is well known
that the landing of a helicopter on a ship deck is very difficult due to the relative motion between the
helicopter and the ship deck on top of the underactuated nature of the helicopter control. A vision system
is considered ideal due to the passive nature of sensing. In this work, a unique marker is conceived to
provide the visual cue for relative attitude and distance. The helicopter uses its onboard camera and
vision system to estimate the relative distance and rotation. The estimation is used to generate the
landing profile.
Since the landing is always relative to the target location, which can be stationary
or moving, it is more important to be able to estimate the relative distance
Related papers:
O. Shakernia, C. S. Sharp, R. Vidal, Y. Ma, S. Sastry, "Multiple View Motion Estimation
and Control for Landing an Unmanned Aerial Vehicle," International Conference on Robotics and Automation,
2002.
Sponsor: Army Research Office (MURI)
The goal of this research lies in the organization of multiple
autonomous agents into organic intelligent systems with minimal supervision and cognition complexity,
high-level of fault-tolerance, and maximized adaptivity to the changes in the task and the environment.
The coordination of multi-agent system is architected as a set of distributed hierarchical hybrid systems,
which emphasize the autonomy of each agent yet allows for coordinated team efforts. The proposed
hierarchical structure transforms a highly complex large scale operation into a set of integrated and
interactive modular functionals. For the validation of the effectiveness of the proposed framework, a
multi-agent probabilistic pursuit-evasion game (PEG) between teams of UAVs and UGVs is chosen. Multi-agent
PEG is a promising benchmark application for cooperative multi-robot systems in which a team of agents
acting as pursers attempts to capture a group of evaders within a bounded but unknown environment. Since
the full solution is often computationally infeasible in realistic situations, a number of suboptimal
approaches are proposed: greedy policy and global-max policy. These proposed ideas were studied in
simulations and compared in full experiments with Berkeley UAVs and UGVs.
Related papers:
R. Vidal, O. Shakernia, H. J. Kim, H. Shim, S. Sastry, "Multi-Agent Probabilistic Pursuit-Evasion Games
with Unmanned Ground and Aerial Vehicles," IEEE Transactions on Robotics and Automation, vol.18, no.5, pp.
662-669, October 2002
H. J. Kim, R. Vidal, D. H.
Shim, O. Shakernia, S. Sastry, "A Hierarchical Approach to Probabilistic Pursuit-Evasion Games with
Unmanned Ground and Aerial Vehicles," IEEE Conference on Decision and Control, 2001
Click here to play a movie: