We present a low power probabilistic floating point multiplier. Probabilistic computation has been shown to be technique for achieving energy efficient designs. As best known the authors, this is first attempt use digital logic attain in To validate approach, multiplications are introduced ray tracing algorithm used computer graphics applications. It then that savings of around 31% can achieved algorithm's multipliers with negligible degradation perceptual quality generated image.

To autonomously move and operate objects in cluttered indoor environments, a service robot requires the ability of 3D scene perception. Though object detection can provide an object-level environmental description to fill this gap, always encounters incomplete observation, recurring detections same object, error detection, or intersection between when conducting continuously room. solve these problems, we propose two-stage algorithm which is fuse multiple views point clouds first stage...

Environmental perception is a vital feature for service robots when working in an indoor environment long time. The general 3D reconstruction low-level geometric information description that cannot convey semantics. In contrast, higher level similar to humans requires more abstract concepts, such as objects and scenes. Moreover, the 2D object detection based on images always fails provide actual position size of object, which quite important robot's operation. this paper, we focus regress...

This paper investigates the use of Reinforcement Learning (RL) to train an Artificial Intelligence (AI) humanoid opponent play a Virtual Reality (VR) table tennis game. A self-play RL algorithm is implemented AI through competing against itself in Unity environment. The simulation environment replicates key physics tennis, and agent controls racket movements hit ball. Experimental result indicates that progressively enhances its skills according rising ELO rating optimization aggressive...

In order to achieve an intuitive interaction and visual semantic navigation for the indoor robot, we propose a novel object-aware hybrid map. The existing map is usually metric map, lacking semantics interaction. We combine objects in environment with constitute consists of 3D object 2D occupancy grid which transfers human commands through semantics, thereby enabling autonomous robot. utilize ORB-SLAM2 continuous pose estimation mapping. detection key-frames conducted based on YOLO v3. point...

Convolutional neural network (CNN)-based methods, which train an end-to-end model to regress a six degree of freedom (DoF) pose robot from single red-green-blue (RGB) image, have been developed overcome the poor robustness visual relocalization recently.However, precision becomes low when test image is dissimilar training images.In this paper, we propose novel method, named image-similarity-based CNN, considers similarity input during CNN training.The higher can achieve.Therefore, crop into...

In this paper, we present our work-in-progress of UniverseSITy, a novel virtual campus which is envisioned to transform applied learning by enabling various computational backend modules. This feature enables UniverseSity be more flexible than the existing education metaverses have heavily-prescribed content or structure.

In this paper, we present an approach for out-of-core dynamic streaming of a virtual scene. We use the traditional client/server architecture, where main responsibility client application is visualization and interaction, server accepting serving requests from application. The will query geometry in camera proximity to stream (in CityGML format) implement using Unity game engine. Performance comparison between loading our are provided. also show scalability advantage work.

We present an efficient resource scheduling scheme for out-of-core dynamic streaming of a 3D scene. The entire scene is stored in the cloud and relevant data are streamed to client mobile device real time based on user-selected path analyze order yield urban objects. compare only user's current location (i.e. without lookahead) terms number loaded objects, rendering performance, storage. application implemented Unity game engine we perform experiment Android device.

We propose a scalable AR (Augmented Reality) multiplayer robotic platform, which enables multiple players to control different machines (a drone and robot) in shared environments, i.e virtual real environments. use state-of-theart visual SLAM (Simultaneous Localization Mapping) algorithms for tracking machine poses based on camera IMU (Inertial Measurement Units) inputs. Players will observe consistent objects between them thanks our backend system, synchronizes the players. Moreover, system...

We present a pilot study on expressive B-spline curves (XBSC), an extension of disk (DBSC). XBSC facilitates drawings in terms shape and color. For shape, colors both sides strokes are defined independently instead using single parameter for as DBSC. perform coloring by considering the envelopes diffusion curves. Our results show that can be used to easily draw wide range images with fewer number primitives compared previous methods.

eXpressive B-Spline Curve (XBSC) is a resolution-independent and computationally efficient technique for vector-based stroke modeling rendering with the flexibility in defining adjusting shape other parameters of stroke. It generalizes existing Disk (DBSC) geometric representation, which itself generalization Bézier curve. XBSC allows flexible color manipulation strokes asymmetrical control rich management. These properties make suitable freeform shapes animation, specifically squash...

We propose CAVRoom, a multiplatform virtual shared space framework for multiplayer sandbox interaction. Participants using various devices such as PCs, Macs, mobile tablets, and phones can interact together in the space. CAVRoom aims to provide accessible solution enhanced remote interactions beyond services offered by existing videoconference solutions. demonstrate classroom setting which students play Heart Attack card game.