ABSTRACT. LiDAR sensors have been very popular in robotics due to their ability to provide accurate range measurements and their robustness to lighting conditions. However, their sensitivity to airborne particles such as dust or fog can lead to perception algorithm failures (e.g. the detection of false obstacles by field robots). In this work, we address this problem by proposing methods to classify airborne particles in LiDAR data. We propose and compare two deep learning approaches, the first is based on voxel-wise classification, while the second is based on point-wise classification. We also study the impact of different combinations of input features extracted from LiDAR data, including the use of multi-echo returns as a classification feature. We evaluate the performance of the proposed methods on a realistic dataset with the presence of fog and dust particles in outdoor scenes. We achieve an F1 score of 94% for the classification of airborne particles in LiDAR point clouds, thereby significantly outperforming the state-of-the-art. We show the practical significance of this work on two real-world use cases: a relative pose estimation task using point cloud matching, and an obstacle detection task. The code and dataset used for this work are available online.

ABSTRACT. By installing omnidirectional measurement light detection and ranging (LiDAR) on a mobile robot, it is possible to acquire the surrounding environment. However, not all measurement areas of LiDAR can be used because of laser rays blocked by the robot's body. Hence, in this research, the authors aim to expand the measurement area of LiDAR by installing mirrors at the blocked area to improve an ability of acquiring information regarding the front environment. The installation position and mirror angle can be determined from the desired area to be measured and the geometrical condition of the laser rays reflecting on the mirror. Moreover, based on the method, the robot enabled to collection of the surface information of stairs during stair climbing, which was typically difficult to gather.

ABSTRACT. The observation of volcanoes is critical in the estimation of volcanic activities and disaster prevention. Volcanic gas from fumaroles is an effective target for these observations. This is because the gas is highly mobile and quickly transmits underground information to the surface. However, as typical volcanic gas is high temperature and contains toxic chemical components, the manned observation of volcanic fumaroles is dangerous. Therefore, in this research, the authors propose an improved version of the dual-wheeled robot to obtain volcanic information from around fumaroles. This robot has a passive tether-guide module to improve its turning range on a slope, which enables the robot to change its traversal direction independent of the tether direction; thus, it has improved turning motion, diagonal traversal motion, and crossing motion on a steep slope. Indoor and field experiments proved the validity of these mechanisms. The lessons learned from these field experiments are also discussed.

ABSTRACT. When a tracked robot moves on a volcanic environment, the robot often has to climb over unfixed obstacles like rocks on the loose ground. On the other hand, tracked robots with subtracks have been proposed, and the climbing performance on fixed obstacles can be improved by controlling the subtracks optimally. However, the effect of subtracks on climbing over the unfixed obstacles has not been clear. In this study, to improve the climbing performance of tracked robots for unfixed obstacles, the effect of subtracks on unfixed obstacles is investigated. Concretely, the climbing-over conditions that a tracked robot should meet in climbing over the unfixed cylindrical obstacle were derived. As a result of the experiment with a real robot, the derived climbing-over conditions are valid, and the climbing over capability of the tracked robot can be increased by setting the subtrack angle optimally. Furthermore, the motion strategies of subtracks are discussed based on the experimental results.

ABSTRACT. While most autonomous driving efforts reported are directed for general driving and mainly on major roads, there are numerous applications for autonomous vehicles for last mile mobility–from person mobility and mail delivery to flexible recharging of cars in parking structures. Over the last year, we have designed vehicles for the micro-mobility challenge. Our approach was based on adoption of the open source Autoware system. The system was taken as a starting point for the design of a robust solution. Proposed requirements include a robust control design, a shift towards increased use of image data over LiDAR data, handling of a richer set of vehicles / pedestrians in a last mile scenario, and overall system characterization and evaluation. We present an overview of the overall design and the design decisions for construction of a vehicles for last-mile delivery.

ABSTRACT. This paper describes an off-line methodology for machines/robots to identify zebra crossings (i.e. for street crossing ability) and their orientations within pedestrian environments. Not knowing crossing ability beforehand can prevent motion trajectories from being accurately planned pre-navigation. As such, we propose a methodology that sources information from 2D maps from the internet to identify the locations of pedestrian zebra crossings. This information is comprised of road networks and satellite imagery of street intersections, from which the locations and orientations of zebra crossings can be identified by means of trained neural networks and proposed verification algorithms. The methodology demonstrated good capability in detecting zebra crossings' locations and verifying them against falsely detected objects in the map. On the other hand, orientation estimation has shown good accuracy, but is yet to be improved.

ABSTRACT. While many researchers have built service robot prototypes that work perfectly under close human supervision, deploying an autonomous robot in an open environment for a long time is not always trivial. This paper presents our experience with TritonBot, a long-term autonomous receptionist and tour guide robot. We used TritonBot as an example to study reliability challenges in long-term autonomous service robots. During the past two years, we regularly do maintenance, fix issues, and roll out new features. In the process, we identified reliability engineering challenges in three aspects of long-term autonomy: scalability, resilience, and learning; we also formulated techniques to confront these challenges. Our experience shows that proper engineering practices and design principles reduces manual interventions and increases general reliability in long-term autonomous service robot deployments.

ABSTRACT. Precise robotic grasping is important for many industrial applications, such as assembly and palletizing, where the location of the object needs to be controlled and known. However, achieving precise grasps is challenging due to noise in sensing and control, as well as unknown object properties. We propose a method to plan robotic grasps that are both robust and precise by training two convolutional neural networks - one to predict the robustness of a grasp and another to predict a distribution of post-grasp object displacements. Our networks are trained with depth images in simulation on a dataset of over 1000 industrial parts and were successfully deployed on a real robot without having to be further fine-tuned. The proposed displacement estimator achieves a mean prediction errors of 0.68cm and 3.42deg on novel objects in real world experiments.

ABSTRACT. In recent years large autonomous vehicles have become more common and especially for outdoor applications often feature four steerable wheels. While these allow the robot to perform more complex maneuvers than, e.g., Ackerman-drive vehicles, such maneuvers can require large changes in the steering angles that are only performed when standing to prevent physical damage to the robot due to mechanical stress. To provide a smooth path execution a local planner that computes velocity commands to follow a given path has to incorporate steering constraints to prevent the robot from stopping unnecessarily to turn its wheels. In this paper we introduce such a planner and two novel methods to incorporate steering constraints. Both are based on considering the instantaneous center of rotation (ICR) and allow only smooth ICR transitions. Extensive experiments in simulation and on the real robot show that our methods notable reduce standing times due to wheel reconfiguration while still accurately following a given path.

ABSTRACT. To achieve high accuracy tracking performance for wheeled mobile robots in spatially varying terrain conditions, it is necessary to estimate both the robot's state, and the slip conditions of the environment, to a high degree of precision. The receding horizon estimation and control (RHEC) framework presents a systematic, adaptive optimisation approach to this problem. In our prior work, the limitations of existing RHEC strategies were outlined, and a structured blocking (SB) method was proposed to address these shortcomings. This was demonstrated in simulation to reduce tracking errors by up to 9% relative to the pre-existing RHEC strategies. In this work, we validate these results in a series of preliminary field experiments with the Swagbot platform, and demonstrate the performance of this SB method. Current and future work will involve performing more comprehensive tests to further characterise and validate control performance.

ABSTRACT. This paper presents the development a four-legged, four-wheel-drive hydraulic powered rover that can traverse irregular terrain such as agricultural or forest landscapes while carrying manipulators etc. to do heavy-duty work. The rover is equipped with driving wheels at the tip of its legs and passive wheels at its knees or the bottom of its belly. By controlling the joints' torque directly using hydraulic actuators, the contact forces at its rolling contact points can be optimally distributed according to the rover's states. This allows the acceleration of the rover's center of gravity and posture to be controlled while accommodating the terrain. This paper describes the mechanical design and control system of the rover, as well as the preliminary experiments. Although the torque control for the driving wheels are not yet successfully implemented, the experimental results using two rover prototypes imply the promising technological capacity of torque-controlled rovers with low-cost hydraulic components.

ABSTRACT. In recent years the range of robotics platforms available for research and development has increased dramatically. Despite this there are areas and applications which are not currently well served by the existing available platforms. Many of them are designed for indoor use; the range of outdoor and off-road robotics platforms is less diverse and few address the issue of deployment in hazardous weather conditions and integrate a suitable sensor suite. In addition almost all of the commercially available Unmanned Ground Vehicles (UGVs) are unsuitable for deployment on delicate surfaces. Given the widespread use of manicured grass within the built environment and agriculture across the western world, this severely limits where they can be deployed and the tasks that can be accomplished. This paper introduces the design principles of a suitable autonomous vehicle. Hulk is built from a commercial zero-turn mower modified for fly-by-wire operation and equipped with a full sensor suite and computing payload. To enable remote, long-term autonomy in diverse environments, several layers of redundant safety systems were designed and installed and the entire assembly made weatherproof.

ABSTRACT. In Japan, the amount of agricultural production decreases year by year. Moreover, reduction in agricultural work population and increase of abandonment of culti-vated land are major social issues. To overcome these issues, we have proposed a small agricultural robot "MY DONKEY" which supports transportation of harvest-ed crops and records the farm work of the user and crop yield in field map while moving closer to the user. In order to move in the furrowed field, it is necessary to detect ridges and furrows where the robot can move using robot mounted sensors and follow the ridge while avoiding the ridges and crop rows. Furthermore, to real-ize smooth harvesting support, we propose a ridge tracking control according to user's behavior based on the recognition of the work contents of the user such as harvesting, loading of harvested crops to the robot, transportation and movement. We propose the ridge tracking control framework based on fuzzy set theory which can evaluate and integrate multiple situations and carry out experiments in straw-berry farm.

ABSTRACT. Forestry is a major industry in many parts of the world. It relies on forest inventory, which consists of measuring tree attributes. We propose to use 3D map- ping, based on the iterative closest point algorithm, to automatically measure tree diameters in forests from mobile robot observations. While previous studies showed the potential for such technology, they lacked a rigorous analysis of diameter estimation methods in challenging forest environments. Here, we validated multiple diameter estimation methods, including two novel ones, in a new varied dataset of four different forest sites, 11 trajectories, totaling 1458 tree observations and 1.4 hectares. We provide recommendations for the deployment of mobile robots in a forestry context. We conclude that our mapping method is usable in the context of automated forest inventory, with our best method yielding a root mean square error of 3.45 cm for our whole dataset, and 2.04 cm in ideal conditions consisting of mature forest with well spaced trees.

ABSTRACT. To help future mobile agents plan their movement in harsh environments,a predictive model has been designed to determine what areas would be favorablefor Global Navigation Satellite System (GNSS) positioning. The model is able topredict the number of viable satellites for a GNSS receiver, based on a 3D pointcloud map and a satellite constellation. Both occlusion and absorption effects ofthe environment are considered. A rugged mobile platform was designed to collectdata in order to generate the point cloud maps. It was deployed during the Canadianwinter known for large amounts of snow and extremely low temperatures. The testenvironments include a highly dense boreal forest and a university campus compris-ing high buildings. The experiment results indicate that the model performs well inboth structured and unstructured environments.

ABSTRACT. The ability to map challenging sub-arctic environments opens new horizons for robotic deployments in industries such as forestry, surveillance, and open-pit mining. In this paper, we explore possibilities of large-scale lidar mapping in a boreal forest. Computational and sensory requirements with regards to contemporary hardware are considered as well. The lidar mapping is often based on the SLAM technique relying on pose graph optimization, which fuses the Iterative Closest Point (ICP) algorithm, Global Navigation Satellite System (GNSS) positioning, and Inertial Measurement Unit (IMU) measurements. To handle those sensors directly within the ICP minimization process, we propose an alternative approach of embedding external constraints. Furthermore, a novel formulation of a cost function is presented and cast into the problem of handling uncertainties from GNSS and lidar points. To test our approach, we acquired a large-scale dataset in the Foret Montmorency research forest. We report on the technical problems faced during our winter deployments aiming at building 3D maps using our new cost function. Those maps demonstrate both global and local consistency over 4.1km.