• 한국HCI학회:학술대회논문집
• /
• 2009.02a
• /
• pp.363-367
• /
• 2009

We propose Cubical User Interface(CUI) for toy block composition in Augmented Reality. The creation of new object by composing virtual object is able to construct various AR contents effectively. However, existing GUI method requires learning time or is lacking of intuitiveness between act of user and offered interface. In case of AR interfaces, they mainly have been supported one handed operation and it did not consider composition property well. Therefore, the CUI provide tangible cube as the manipulation tool for virtual toy block composition in AR. The tangible cube which is attached multi-markers, magnets, and buttons supports free rotation, combination, and button input. Also, we propose two kinds of two-handed composing interactions based on CUI. First is Screw Driving(SD) method which is possible to free 3-D positioning and second is Block Assembly(BA) method which support visual guidance and is fast and intuitive. We expected that proposed interface can apply as the authoring system for content such as education, entertainment, Digilogbook.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.3
• /
• pp.141-148
• /
• 2022

Recently, reinforcement learning combined with deep neural network technology has achieved remarkable success in various fields such as board games such as Go and chess, computer games such as Atari and StartCraft, and robot object manipulation tasks. However, such deep reinforcement learning describes states, actions, and policies in vector representation. Therefore, the existing deep reinforcement learning has some limitations in generality and interpretability of the learned policy, and it is difficult to effectively incorporate domain knowledge into policy learning. On the other hand, dNL-RRL, a new relational reinforcement learning framework proposed to solve these problems, uses a kind of vector representation for sensor input data and lower-level motion control as in the existing deep reinforcement learning. However, for states, actions, and learned policies, It uses a relational representation with logic predicates and rules. In this paper, we present dNL-RRL-based policy learning for transportation mobile robots in a manufacturing environment. In particular, this study proposes a effective method to utilize the prior domain knowledge of human experts to improve the efficiency of relational reinforcement learning. Through various experiments, we demonstrate the performance improvement of the relational reinforcement learning by using domain knowledge as proposed in this paper.

• Journal of the Korea Computer Graphics Society
• /
• v.3 no.1
• /
• pp.57-67
• /
• 1997

The navigation of virtual space comes down to the manipulation of the virtual camera. The movement of the virtual cameras has 6 degrees of freedom. However, input devices such as mouses and joysticks are 2D. So, the movement of the camera that corresponds to the input device is 2D movement at the given moment. Therefore, the 3D movement of the camera can be implemented by means of the combination of 2D and 1D movements of the camera. Many of the virtual space navigation browser use several navigation modes to solve this problem. But, the criteria for distinguishing different modes are not clear, somed of the manipulations in each mode are repeated in other modes, and the kinesthetic correspondence of the input devices is often confusing. Hence the user has difficulty in making correct decisions when navigating the virtual space. To solve this problem, we use a single navigation metaphore in which different modes are organically integrated. In this paper we propose a helicopter pilot metaphor. Using the helicopter pilot metaphore means that the user navigates the virtual space like a pilot of a helicopter flying in space. In this paper, we distinguished six 2D movement spaces of the helicopter: (1) the movement on the horizontal plane, (2) the movement on the vertical plane,k (3) the pitch and yaw rotations about the current position, (4) the roll and pitch rotations about the current position, (5) the horizontal and vertical turning, and (6) the rotation about the target object. The six 3D movement spaces are visualized and displayed as a sequence of auxiliary windows. The user can select the desired movement space simply by jumping from one window to another. The user can select the desired movement by looking at the displaced 2D movement spaces. The movement of the camera in each movement space is controlled by the usual movements of the joystick.