• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2002년도 춘계학술대회논문집
• /
• pp.842-847
• /
• 2002

One of the challenging problems with bicycle simulators is to deal with the inherent unstable bicycle dynamics that is coupled with rider's motion. For the bicycle dynamics calculation and the real time simulation, it is necessary to identify the control inputs from the rider as well as the virtual environments. The six control forces of the Stewart platform-based motion system are used for estimation of the rider's action force, which is one of the important control inputs, but of which the direct measurement is impractical. For the effective estimation of the rider's action force, the dynamics model of the motion system is derived incorporated with both analytical and experimental methods and the sliding mode controller with perturbation estimation is developed.

• 대한기계학회논문집A
• /
• 제36권11호
• /
• pp.1319-1326
• /
• 2012

본 논문은 구동력이 자전거의 self-stability 에 미치는 영향을 해석한다. 운전자의 제어를 고려하지 않는 자전거의 self-stability 는 자전거의 선형 모델로부터 고유치를 구하여 해석 할 수 있다. 전륜과 후륜에 작용하는 자전거 구동력을 고려한 선형모델을 개발하고 이로부터 구동력이 안정성에 미치는 영향을 규명하였다. 선형모델의 결과가 다물체 동역학 모델의 시뮬레이션을 통한 비선형 자전거 모델에서도 같은 결과를 보임을 확인하였다.

• 제어로봇시스템학회논문지
• /
• 제20권5호
• /
• pp.551-556
• /
• 2014

This paper proposes a balancing control and driving control of a bicycle robot based on dynamic modeling of the bicycle robot, which has been derived using the Lagrange equations. For the balancing control of the bicycle robot, a reaction wheel pendulum method has been adopted in this research. By using the dynamics equations of the bicycle robot, an LQR controller has been designed for a balancing and driving control of a bicycle robot. The performance of the balance control is verified experimentally before the driving control, which shows a stable posture within one degree vibrations. To show the dynamic characteristics of the bicycle robot during driving, a trapezoidal velocity trajectory is selected as the references. Through simulations and real experiments, the effectiveness of the proposed algorithm has been demonstrated.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2001년도 ICCAS
• /
• pp.117.5-117
• /
• 2001

This paper presents the key idea of handlebar reaction force and pedal resistance force generation in creating life-like feeling in KAIST bicycle simulator. Also, it provides methods to evaluate its reality level with given reaction force profile. In KAIST bicycle simulator, the pedal resistance force and the handlebar reaction force are calculated using the bicycle dynamic model. With the information handlebar angle, rider´s pedaling torque and road profile transmitted from the handlebar system, the pedal system and the visual part, the bicycle dynamics engine calculates the handlebar reaction force and the pedal velocity. The handlebar system and the pedal resistance system generate reaction force and resistance force transmitted from dynamics engine. However to make more realistic riding feeling ...

• Journal of Mechanical Science and Technology
• /
• 제19권spc1호
• /
• pp.292-304
• /
• 2005

In this paper we present the linearized equations of motion for a bicycle as a benchmark. The results obtained by pencil-and-paper and two programs are compared. The bicycle model we consider here consists of four rigid bodies, viz. a rear frame, a front frame being the front fork and handlebar assembly, a rear wheel and a front wheel, which are connected by revolute joints. The contact between the knife-edge wheels and the flat level surface is modelled by holonomic constraints in the normal direction and by non-holonomic constraints in the longitudinal and lateral direction. The rider is rigidly attached to the rear frame with hands free from the handlebar. This system has three degrees of freedom, the roll, the steer, and the forward speed. For the benchmark we consider the linearized equations for small perturbations of the upright steady forward motion. The entries of the matrices of these equations form the basis for comparison. Three diffrent kinds of methods to obtain the results are compared : pencil-and-paper, the numeric multibody dynamics program SPACAR, and the symbolic software system Auto Sim. Because the results of the three methods are the same within the machine round-off error, we assume that the results are correct and can be used as a bicycle dynamics benchmark.

• 제어로봇시스템학회논문지
• /
• 제17권10호
• /
• pp.1006-1013
• /
• 2011

In this paper, we propose an optimal posture control law for an unmanned bicycle by deriving linear bicycle model from fully nonlinear differential equations. We calculate each equilibrium point of a bicycle under any given turning radius and angular speed of rear wheel. There is only one equilibrium point when a bicycle goes straight, while there are a lot of equilibrium points in case of turning. We present an optimal equilibrium point which makes the leaning input minimum when a bicycle is turning. As human riders give rolling torque by moving center of gravity of a body, many previous studies use a movable mass to move center of gravity like humans do. Instead we propose a propeller as a new leaning input which generates rolling torque. The propeller thrust input makes bicycle model simpler and removes input magnitude constraint unlike a movable mass. The proposed controller can hold optimal equilibrium points using both steering input and leaning input. The simulation results on linear control for circular motion are demonstrated to show the validity of the proposed approach.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2001년도 추계학술대회논문집 I
• /
• pp.235-240
• /
• 2001

A four degrees of freedom (dof) motion platform for bicycle simulator is developed. The motion platform, capable of the vertical linear and three angular motions, is designed based on analysis of the typical motion characteristics revealed by the existing six dof bicycle simulator. The platform essentially consists of two parts： the three dof parallel manipulator, consisting of a moving platform, a fixed base and three actuators, and the turntable to generate the yaw motion. The nonlinear kinematics and dynamics of the three dof parallel manipulator with multiple closed loop chains are analyzed for tracking control of the motion platform. The tracking performances of the three control schemes are experimentally compared： the computed torque method (CTM), the sliding mode control (SMC) and the PD control. The CTM and SMC, incorporated with the system dynamics model, are found to be equally better in performance than the PD controller, irrespective of the presence of external disturbance.

• 한국컴퓨터정보학회논문지
• /
• 제23권10호
• /
• pp.23-31
• /
• 2018

Recently, there have been many studies on machine learning. Among them, studies on reinforcement learning are actively worked. In this study, we propose a controller to control bicycle using DDPG (Deep Deterministic Policy Gradient) algorithm which is the latest deep reinforcement learning method. In this paper, we redefine the compensation function of bicycle dynamics and neural network to learn agents. When using the proposed method for data learning and control, it is possible to perform the function of not allowing the bicycle to fall over and reach the further given destination unlike the existing method. For the performance evaluation, we have experimented that the proposed algorithm works in various environments such as fixed speed, random, target point, and not determined. Finally, as a result, it is confirmed that the proposed algorithm shows better performance than the conventional neural network algorithms NAF and PPO.

• Journal of Positioning, Navigation, and Timing
• /
• 제11권3호
• /
• pp.209-215
• /
• 2022

As intelligent autonomous driving vehicle development has become a big topic around the world, accurate reference dynamics estimation has been more important than before. Current systems generally use speed and heading information sensed from a distant point as a vehicle reference dynamic, however, the dynamics between different points are not same especially during rotating motions. In order to estimate properly estimate the reference dynamics from the information such as velocity and heading sensed at a point distant from the reference point such as center of gravity, this study proposes estimating reference dynamics from any location in the vehicle by combining the Bicycle and Ackermann models. A test system was constructed by implementing multiple GNSS/INS equipment on an Robot Operating System (ROS) and an actual car. Angle and speed errors of 10° and 0.2 m/s have been reduced to 0.2° and 0.06 m/s after applying the suggested method.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2011년도 추계학술대회
• /
• pp.603-604
• /
• 2011

자전거 운전은 본질적인 불안정한 시스템으로 자전거의 안전한 운전을 위해서는 지속적인 운전자의 상호작용으로 인하여 자전거의 안전성을 확보할 수 있다. 그것은 자전거의 운전이 운전자의 운동과 매우 밀접한 연관이 있다는 것을 알 수 있으며, 현재 자전거의 특성뿐 아니라 운전자의 운전에 대한 연구가 진행되고 있지만 정확한 자전거의 주행 안정성에 대하여 연구는 계속적으로 진행되고 있다. 본 논문에서는 자전거 운전자의 몸무게와 신장과 같은 사람의 인체 정보를 이용하여 적절한 안장 높이와 자전거 바퀴 사이의 거리 및 핸들의 높이를 조절할 수 있는 맞춤형 자전거 관리 시스템을 제시하고자 한다. 또한, 운전자 맞춤형 자전거의 설계는 실제 자전거 운전데이터와 영상데이터를 이용하여 적합한 자전거 모델을 제시하고 추출된 데이터를 통하여 분석된 결과를 제시하고자 한다.