Publications

Autonomous vehicles require high-level semantic maps, which contain the activities of pedestrians and cars, to ensure safe navigation. High-level semantics can be obtained from mobile probe sensor data. Analyzing pedestrian trajectories obtained from mobile probe data is an effective approach to avoid collisions between autonomous vehicles and pedestrians. Such analyses of pedestrian trajectories can generate new information such as pedestrian behaviors in violation of traffic regulations. However, pedestrian trajectories obtained from mobile probe data significantly sparse and noisy, making it challenging to analyze pedestrian activity. To address this issue, we propose multiple daily data and graph-based approaches to treat sparse and noisy data for estimating the flow of pedestrians based on mobile probe data. To improve the sparseness of the data, multiple daily data are fused. After that, a pedestrian graph is created to enhance the region’s coverage by connecting the sparse data indicating the flow of pedestrians. This proposed approach successfully obtained pedestrian trajectory data from the sparse and noisy data. Moreover, it was possible to identify the potential locations where pedestrians tend to cross the street by analyzing the pedestrian flow. The results indicate that 83% of well-known regions where pedestrians tend to cross the street corresponded with those extracted using the proposed approach. Furthermore, a high-level semantic map of the regions where pedestrians tend to cross the street along a 1-km road is presented. The trajectory information obtained using the proposed approach is expected to be essential for understanding different scenarios of the interactions between individuals and autonomous vehicles.

Automated Driving Systems (ADSs) are being proposed as a promising technology that will help drivers avoid accidents, as well as reduce driving-related stress. To enhance the driving performance of ADSs, driving log data from human drivers is being used to teach these systems how to behave in different traffic regions. Visual information from the driving log data is an easy method to identify traffic regions. However, due to space constraints or lack of visual sensors, most of the driving log data has no visual information. Therefore, another solution is necessary to identify traffic regions using only spatial information. To address this challenge, the paper proposes a region recognition method using key primitive transitions obtained from vehicle trajectory information. To obtain the motion primitives, we employed a hierarchical similarity clustering that combines hierarchical clustering and HDP-HSMM. The vehicle behavior from the region of interest was extracted by analyzing the motion primitives transitions located within close range (50 m) of the region. We assessed the behavior of vehicles when approaching two regions: traffic lights and entrances. Three environments were used to evaluate the proposed method, with two different drivers and distinguished layouts. This study shows that the proposed classification method can identify traffic light and entrance regions with an average 89% precision and 86% f-score. Additionally, the hierarchical similarity clustering and window observation approach developed in this study is responsible for increasing the system's precision by 9%.

Semantic maps are an important tool to provide robots with high-level knowledge about the environment, enabling them to better react to and interact with their surroundings. However, as a single measurement of the environment is solely a snapshot of a specific time, it does not necessarily reflect the underlying semantics. In this work, we propose a method to create a semantic map of a construction site by fusing multiple daily data. The construction site is measured by an unmanned aerial vehicle (UAV) equipped with a LiDAR. We extract clusters above ground level from the measurements and classify them using either a random forest or a deep learning based classifier. Furthermore, we combine the classification results of several measurements to generalize the classification of the single measurements and create a general semantic map of the working site. We measured two construction fields for our evaluation. The classification models can achieve an average intersection over union (IoU) score of 69.2 during classification on the Sanbongi field, which is used for training, validation and testing and an IoU score of 49.16 on a hold-out testing field. In a final step, we show how the semantic map can be employed to suggest a parking spot for a dump truck, and in addition, show that the semantic map can be utilized to improve path planning inside the construction site.

Sensing position and orientation of construction vehicle is an important issue for automation of construction process. We aim to develop sensing and visualization technologies for construction vehicles. Our target construction vehicle is backhoes. Construction vehicles are usually rented in construction fields. However, construction vehicles that can be rented do not have the sensing function. It is too hard to obtain backhoe position and manipulator pose without sensing information. This paper proposes an attachable sensor box that can measure backhoe position and orientation and the manipulator pose. The sensor boxes can be attached on backhoe metal surface by magnetic force without any additional manufacturing on the construction vehicle surface. By using WiFi communication and mobile battery the sensor box can be easily attached on large-size backhoe without wiring. After the work is done, it is easily detached by changing magnetic force power. At loading and scooping motion, backhoe arm and boom have large force exceeding 16 G. To attach the sensor boxes to the arm and boom, we designed an additional magnetic frame that can generate force of 1960 N. The sensor boxes were firmly attached to each joint and prevented to drop by any force from backhoes movement. We also measured the behavior of the backhoe in loading works in hot conditions over 30 °C and visualized backhoe pose and tip manipulator position.

Backhoe loads sediment onto the bed of dump trucks during earthmoving work. The prediction of backhoe loading time is essential for ensuring safe cooperation between the backhoe and dump trucks. However, it is difficult to predict the instant at which the backhoe is ready to load sediment, because of the similarity in motions observed during gathering sediment. Moreover, since operators have different skill levels, the prediction requires a unique model for each operator. In this study, we attempt to predict the instant at which the backhoe is ready to load sediment into the dump truck. For this purpose, the beta-process hidden Markov model (BP-HMM) is employed to build a backhoe motion model for a specific operator. Time series data of backhoe loading motions for crushed rocks and wood chips, which were measured using 6-axis inertial measurement unit (IMU) sensors equipped at the cab, boom, and arm of the backhoe, were used for modeling with the BP-HMM. Several primitive motions of the backhoe, which occur at the completion of preparation before the loading process begins, were discovered as a result of the motion modeling based on the BP-HMM. We developed the prediction of the instant using three primitive motions. At best, the proposed method could predict the instant with a probability of 67% and 100%, at 6.0 s and 0.7 s before the loading motions began, respectively. This phased prediction can be used to reduce the idle time and risk for dump trucks during earthmoving work with the backhoe.

The proposal of this work consists in an approach based on the utilization of two Wi-Fi receivers, using RSS information received by the transmission nodes (Access Point) for the SLAM problem. The solution presented is based on extended Kalman filter using RSS coming from different transmitters in the environment. This approach was chosen because the Cellbot platform has two Wi-Fi receivers, therefore combining the two range-only sensors improves the precision of the system. The system considers that a robot navigates in an unknown environment where it receives different Wi-Fi signals from a known source. Through the odometry and Wi-Fi signals, the robot localizes itself and the position of each Wi-Fi transmitter. The results show the viability of the application, and an increase of 38% in precision of the robot pose by using proposed approach.

One of the major challenges that autonomous cars are facing today is the unpredictability of pedestrian movement in urban environments. Since pedestrian data acquired by vehicles are sparse observed a pedestrian flow directed graph is proposed to understand pedestrian behavior. In this work, an autonomous electric vehicle is employed to gather LiDAR and camera data. Pedestrian tracking information and semantic information from the environment are used with a probabilistic approach to create the graph. In order to refine the graph a set of outlier removal techniques are described. The graph-based pedestrian flow shows an increase of 61.29 % of coverage zone, and the outlier removal approach successfully removed 81 % of the edges.

This paper goal is to present a systematic revision of the works related to robotic on education, which were published on three national events of this area: Informatics on Education Brazilian Symposium (SBIE), Informatics on School Workshop (WIE), Robotics on Education Workshop (WRE), between 2004 to 2014. The results shown an increasing interest of brazilian community for the area, highlighting the northeast and southeast regions of Brazil responsible for most of the works. It was also found that 63% of this works represents proposals of the use of robotics on elementary school with new methodologies and robots to assist pedagogy in the use of robotics.