The unlawful disposal of aluminum and plastic materials, such as empty cans and plastic bottles, has been a serious social issue. We are all concerned that this issue could harm ecosystems and increase pollution. As a result, social volunteer projects have started picking up trash to address this issue. These activities, however, take a lot of time and cost. Therefore, in this work, we propose the automation of trash pickup by robots as a solution to this problem. Trash pickup robots must perform several functions, including moving autonomously and detecting and collecting unlawfully dumped materials. This study develops a technology to detect illegally dumped objects from videos captured by a robot. Furthermore, this research suggests a faster object detection technique that enhances the YOLO low memory, quick method. Additionally, we developed an annotated image dataset with three new types of trash (plastic bottles, cans, and cardboard). We used this data in validation trials to compare the performance of the proposed method with that of existing methods. The experimental results demonstrated that our model could detect the three types of trash faster than the selected models and enhanced the processing speed. Finally, a comparison was made for frames per second on GPUs and edge devices to validate whether all models could detect trash in realtime.
In the construction industry, the reduction in labor population due to the declining birthrate and aging population has huge implications on the overall productivity in the industry, for which the adoption of robotic technologies can be a significant solution in tackling with the challenges. Nevertheless, considering the complexity of the construction process, it is needed to have coordinated operation of various autonomous construction machinery working towards certain common goals. Moreover, the cluttered and unstructured nature of the construction sites requires a sound motion planning method that predicts and recognizes the situation and takes correct decisions. Algorithms for both need to be tested and improved continuously via simulation before real construction applications. This paper proposes the concept of an integrated controller for coordinated operation and motion planning of autonomous construction machinery and its system configuration was discussed. Unity is proposed to be the controller due not only to its seamless communication with the Robot Operating System (ROS) in exchanging control commands via the rosbridge library but also to its capability of importing models of various data formats in creating a high- fidelity simulation environment. With the prototype validation of the proposed framework, a workflow from construction planning, simulation verification, to real machine control is expected to bring huge benefits to the construction industry in terms of improved productivity, quality, and safety.
This study uses an unmanned aircraft mounted green Light Detection and Ranging (LiDAR) system to verify the learnability associated with deep learning regarding land cover classification (i.e., evaluated by the average and the absolute difference value of label-based F1, overall accuracy or Macro F1 under the premise of cross-seasonal mutual-prediction). Image fusion method in this study mainly attempts to superimpose the visualized LiDAR data with the aerial photographs to provide new features. LiDAR data are visualized using high contrast color scale (default), average high contrast color scale (same color, different split points as default), and high contrast gray color scale (different color, same split points as default). The reason of setting these methods is to compare the impact of split points and point-associated color in the color scale on learnability of cross-seasonal data. It is worth noting, however from the view of the results, that relying solely on aerial photographs is not sufficient, comparing with image fusion method, especially when training and predicting using cross-seasonal data. From the comparison of the label-based F1, this study has proved the necessary impact of LiDAR on vegetation label-based land cover classification.
The conventional road crack ratio evaluation in Japan uses top-view images of pavement obtained from precision line scan cameras and manually classifies 0.5m grids of pavement surface based on the number of cracks. While inexpensive crack evaluation using in-vehicle smartphone cameras has been proposed, previous work cannot calculate the crack ratio based on the index definition and have limited accuracy. This paper proposes a vision-based top-view transformation and image stitching algorithm for road crack ratio evaluation using video captured by an in-vehicle smartphone camera. Four conditions are used to perform the parameter calibration: 1) horizontal manhole axis, 2) parallel lane line, 3) circular manhole, 4) vertical lane line conditions. After the successful top-view transformation, feature matching is conducted to pairs of successive frames to calculate the homography matrix between the two images, which is used for the image stitching of successive frames and obtaining the translation offset between the images. Based on the calculated translation offset and the extracted frame distance interval, the pixel-to-real-distance conversion factor is calculated. The image is divided into a 0.5m grid based on the index definition. An image classification model was trained to classify each grid box according to the number of cracks. The results showed that: 1) a fine-resolution image of road top-view can be produced from successive images captured by an in-vehicle smartphone camera, and 2) the crack ratio can be accurately estimated from these images automatically.
Building information modeling/construction information management (BIM/CIM) has been widely adopted as the main technology in civil engineering and related fields. It has enabled remarkable innovations in the construction industry. However, despite its contribution in facilitating construction practices, the application of BIM for detailed design has mainly been dedicated to 3D visualization and clash detection in civil engineering projects; thus, it has not been fully exploited. Therefore, in this study, we developed an automated verification system for structural details to rationalize the creation of rebar shop drawings using BIM. We focused on lap splicing as a structural detail. In particular, we first designed the algorithm required for verification, and then developed a prototype. Further, we verified its effectiveness using a sample model.
With the number of heavy rainfall events on the rise, existing dam facilities must maximize their functions to reduce flood damage. Artificial intelligence (AI) has been applied to improving the efficiency of dam operation during floods, but previous research focused on flood control of a single dam only. In this study, a model to operate multiple dams for flood control was constructed by using the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm for deep reinforcement learning. The dam operation AI was applied to operating three dams in the Kinu River basin against generated rainfall data, and the results were compared with flood control according to dam operation rules. The results showed that the dam operation AI could realize a significantly lower average maximum flow in the middle reaches of the Kinu River than that of flood control according to dam operation rules.
During the 2011 off the Pacific coast of Tohoku Earthquake, many residents living coastal area used car for their evacuation from mega Tsunami. Due to the rapidly increased traffic demand in road network and also trouble of traffic signals, heavy traffic congestions occurred in the road network. At first, this research investigated the condition of car evacuation in Kesennuma city. Secondly, tsunami evacuation simulation system was developed. Thirdly, several counter-measures for safer tsunami evacuation were developed and proposed. Finally, the scenario of roads widening proposed by Kesennuma city was evaluated using the simulation system. As a result, it was clarified that road widening is an effective countermeasure to modify the road traffic flow and that it could not solve the problems in car evacuation perfectly.During the 2011 off the Pacific coast of Tohoku Earthquake, many residents living coastal area used car for their evacuation from mega Tsunami. Due to the rapidly increased traffic demand in road network and also trouble of traffic signals, heavy traffic congestions occurred in the road network. At first, this research investigated the condition of car evacuation in Kesennuma city. Secondly, tsunami evacuation simulation system was developed. Thirdly, several counter-measures for safer tsunami evacuation were developed and proposed. Finally, the scenario of roads widening proposed by Kesennuma city was evaluated using the simulation system. As a result, it was clarified that road widening is an effective countermeasure to modify the road traffic flow and that it could not solve the problems in car evacuation perfectly.