Search

Household chores service is one of the desirable functions for a service robot. Object classification is the most important function when searching for objects. We consider that the major concern is that the robot should not misclassify the object. If the robot misclassifies household object images, it will then perform the household tasks with the wrong object. This may cause serious damages to a service robot and to the user. This concept gives us the insight that the precision of the classification must be very high by setting a confidence threshold so that it then can claim a reliable service robot. By exploring this concept, we develop a more convincing indicator, Classification Reliability, to reveal the reliability of deep learning model. Moreover, we develop a fine-tune rule base to continuously regenerate more proper training dataset for the CNN model to increase reliability. Experimental results demonstrate that the CNN model fine-tuned by our closed-loop system achieves the reliability which is higher than the other similar effects such as DenseNet on the CIFAR-10 dataset.

Image captioning is a high-level task that generates the context scenario descriptions from an image. There are many ways to implement such kinds of ability that can generate a smooth sentence about the context scenario. Because of the open and extensive domain of the dataset, seldom of them can really help people to get the meaningful information of a specified scenario. In this paper, we propose a novel framework called Human-Aware Context Generator (HACG). We adopt the concept of dividing and conquer, which combines the ability of face recognition and facial expression recognition while retaining the capability of image caption model. This model aims to offer a significant context scenario sentence to the people, especially for those who are urgent to master the overall situation.

Image caption generation is a task that generates a sentence from a raw image, which is mimicking the intelligence of human that can acquire knowledge from the view. The difficulty of this task is the combination of multimodal knowledge learning, i.e. recognition of objects, actions, scenes, human, etc. In order to perform semantic understanding for service robotics or other industrial applications, the caption must be enhanced for recognition of the objects in the confined environment. We propose a template-based augmentation method for improving the capability of object recognition while retaining the other capability of the image caption model. This work opens a new era of image caption generation training procedure that the caption dataset and the classification dataset can be combined to train the deep captioning model. We show in our experiments that our improved model outperforms the original model in SPICE metrics by 4 times.

Vehicular Ad Hoc Network (VANET), containing access points free for wireless transmission, is now developed for V2V and V2I communication, in order to improve traffic safety, provide services and further support the implementation of the automated driving mode. However, the most important issue is how to develop an efficient VANET protocol, particularly corresponding to reduce transmission latency when routing. This paper proposes a novel multi-hop routing algorithm to accomplish message transmission via the nodes participating in the given vehicular ad-hoc network. According to the testing result, the proposed protocol is acceptable.

In this work, an effective energy-saving device is proposed to extract wave energy and convert to useful propulsive thrust for a merchant ship travelling in waves. Eco-Ship was internationally introduced and aimed to reduce carbon dioxide emission generated from ships. The energy-saving device equipped underneath the front part of the ship model of general merchant ships that inspired by marine mammals is investigated in this work. An active control of pitch-oscillating bow fin with NACA0016 section is applied as an effective energy-saving device in this work. Free-running model test using a 1/50th scale model of a container ship was conducted for validating the effectiveness of this device. The ship speed, ship motion, angle of bow fin and power consuming are measured and recorded. There are two kinds of wavelength to ship length ratio of regular wave in the experiments with fixed propeller speed. Based on different heave motion, the different pitch angle of bow fin is controlled to obtain the extra energy from wave. The experimental results shows the feasibility of the energy-saving device when the phase lead of the bow fin pitch motion with respect to its heave motion set in a proper range between 60 degree to 200 degree. When the ratios of wavelength to ship length are 1.0 and 1.3, the maximum of power efficiency improving are around 6.5% and 18.9% respectively. The experimental results demonstrated that the proposed device is effective to enhance ship speed and save energy for achieving the goal of Eco-Ship.

Boron doping on the structural, optical and electrical properties in BZO films fabricated by spray pyrolysis were investigated. The carrier concentration of ∼1019cm−3 and optical transmittance about 90% in the infrared region was achieved.

Thin films of Cuprous Oxide (Cu 2 O) with different deposition temperatures have been deposited on glass substrates by spray pyrolysis. The films were investigated in terms of morphological surface, structural and electrical properties exhibited p-type conduction.