ABSTRACT. We here address the issue of air vehicles flying autonomously at a high speed in complex environments. Typically, autonomous navigation through a complex environment requires a continuous heuristic search on a graph generated by a k-connected grid or a probabilistic scheme. The process is expensive especially if the paths must be kino-dynamically feasible. Aimed at tackling the problem from a different angle, we consider the case that the environment is mostly known from a prior map. The proposed method suggests the computation needed to find safe paths during fast flight can be greatly reduced if we pre-compute and carefully arrange a set of alternative paths before the flight. During the navigation, the vehicle selects a pre-computed path to navigate without the need to generate a new path. The result is that majority of the processing is migrated to offline path generation. Effectively, the onboard computation is significantly reduced, taking <3% of a CPU thread on a modern embedded computer. In experiments, it enables a lightweight aerial vehicle to maneuver aggressively through a cluttered forest environment at 10m/s.

ABSTRACT. This paper develops an autonomous tethered aerial visual assistant for robot operations in unstructured or confined environments. Robotic tele-operation in remote environments is difficult due to lack of sufficient situational awareness, mostly caused by the stationary and limited field-of-view and lack of depth perception from the robot's onboard camera. The emerging state of the practice is to use two robots, a primary and a secondary that acts as a visual assistant to overcome the perceptual limitations of the onboard sensors by providing an external viewpoint. However, problems exist when using a tele-operated visual assistant: extra manpower, manually chosen suboptimal viewpoint, and extra teamwork demand between primary and secondary operators. In this work, we use an autonomous tethered aerial visual assistant to replace the secondary robot and operator, reducing human robot ratio from 2:2 to 1:2. This visual assistant is able to autonomously navigate through unstructured or confined spaces in a risk-aware manner, while continuously maintaining good viewpoint quality to increase the primary operator's situational awareness. With the proposed co-robots team, tele-operation missions in nuclear operations, bomb squad, disaster robots, and other domains with novel tasks or highly occluded environments could benefit from reduced manpower and teamwork demand, along with improved visual assistance quality based on trustworthy risk-aware motion in cluttered environments.

ABSTRACT. Planning for multi-robot coverage seeks to determine collision-free pathsfor a fleet of robots, enabling them to collectively observe points of interest in an en-vironment. Persistent coverage is a variant of traditional coverage, where coverage-levels in the environment decay over time. Thus robots have to continuously revisitparts of the environment to maintain a desired coverage-level. Facilitating this in thereal world demands us to tackle numerous subproblems. While there exist standardsolutions for these subproblems, there is no complete framework that addresses allof their individual challenges as a whole in a practical setting. We adapt and com-bine such solutions to present a planning framework for persistent coverage withmultiple unmanned aerial vehicles (UAVs). More specifically, we run a persistentloop of goal-assignment and globally deconflicting, kinodynamic path-planning formultiple UAVs. We evaluate our framework in simulation as well as the real world.Specifically, we demonstrate that (i) our framework exhibits the desirable traits ofgraceful degradation—given sufficient resources, we maintain persistent coverage,whereas while resources decrease (i.e., environment size increases or number ofUAVs decreases) coverage-levels decay slowly and (ii) ensuring global deconflic-tion in our framework incurs a negligibly higher price compared to other, weakercollision-checking schemes.

ABSTRACT. This work presents a system and approach for rapid exploration of unknown environments using aerial robots. High-speed flight with multirotor air vehicles is challenging due to limited sensing range, use of on-board computation, and constrained dynamics. For robots operating in unknown environments, the control system must guarantee collision-free operation, and for exploration tasks, the system should also select sensing actions to maximize information gain with respect to environment. To this end, we present a motion primitive-based, receding-horizon planning approach that maximizes information gain, accounts for platform dynamics, and ensures safe operation. Analysis of motions parallel and perpendicular to frontiers given constraints on sensing and dynamics leads to bounds on safe velocities for exploration. This analysis and the bounds obtained, inform the design of the motion primitive approach. Simulation experiments in a complex 3D environment demonstrate the utility of the motion primitive actions for rapid exploration and provide a comparison to a reduced motion primitive library that is appropriate for online planning. Experimental results on a hexarotor robot with the reduced library demonstrate rapid exploration at speeds above 2.25 m/s under a varying clutter in an outdoor environment which is comparable to and exceeding the existing state-of-the-art results.

ABSTRACT. In this paper, we present a path planner for low-altitude terrain coverage in known environments with unmanned rotary-wing micro aerial vehicles (MAVs). Airborne systems can assist humanitarian demining by surveying suspected hazardous areas with cameras, ground-penetrating synthetic aperture radar, and metal detectors. Most available coverage planner implementations for MAVs do not consider obstacles and thus cannot be deployed in obstructed environments. We describe an open source framework to perform coverage planning in polygon flight corridors with obstacles. Our planner extends boustrophedon coverage planning by optimizing over different sweep combinations to find the optimal sweep path, and considers obstacles during transition flights between cells. We evaluate the path planner on 320 synthetic maps and show that it is able to solve realistic planning instances fast enough to run in the field. The planner achieves 14% lower path costs than a conventional coverage planner. We validate the planner on a real platform where we show low-altitude coverage over a sloped terrain with trees.

ABSTRACT. Among the most challenging of environments in which an autonomousmobile robot might be required to serve is the subterranean environment. The com-plete lack of ambient light, unavailability of GPS, and geometric ambiguity makesubterranean simultaneous localization and mapping (SLAM) exceptionally diffi-cult. While there are many possible solutions to this problem, a visual-inertial frame-work has the potential to be fielded on a variety of robotic platforms which can op-erate in the spatially constrained and hazardous environments presented by the sub-terranean domain. In this work we present an evaluation of visual-inertial SLAMin the subterranean environment with onboard lighting and show that it can con-sistently perform quite well. However, this performance is dependent on includingsome modifications that depart from the typical formulation of VI-SLAM, as wellas careful tuning of the system’s visual tracking parameters. We discuss the some-times counter-intuitive effects of these parameters and provide insight into how theyaffect the system’s overall performance.

ABSTRACT. In this paper a comprehensive approach to enable resilient robotic autonomy in subterranean environments is presented. Emphasizing on the use of aerial robots to explore underground settings such as mines and tunnels, the presented methods address critical challenges related to extreme sensor degradation, path planning in large-scale, multi-branched and geometrically-constrained environments, and reliable operation subject to lack of communications. To facilitate resilience in such conditions, novel methods in multi-modal localization and mapping, as well as graph-based exploration path planning are proposed and combined with custom system design. Through a set of field evaluation activities in real-life subterranean environments we present a "field-hardened" solution that demonstrably enables reliable robotic operation in the hard to access but often crucial underground settings.

ABSTRACT. This paper presents a real-time viable method for Simultaneous Localization and Mapping (SLAM) for compute-constrained systems in subterranean environments using Gaussian mixture models (GMMs). The two contributions of this work are (1) a SLAM formulation that uses a GMM-based map representation for pose estimation, mapping and loop closure and (2) an Expectation Maximization (EM) formulation that significantly reduces the time to train a GMM from a sensor observation by exploiting the insight that although Gaussian distributions have infinite support, a substantial contribution is contained within a finite support region. An on-manifold distribution-to-distribution registration approach is used to estimate pose between consecutive GMMs and the Cauchy-Schwarz divergence is employed to calculate the similarity between the distributions and identify loop closures. The method is evaluated in an unstructured cave environment. The results demonstrate significant accuracy advantages to leveraging the compact representation of the GMM as compared to traditional pose graph SLAM techniques that rely on pointcloud-based methods, particularly when closing the loop. Further, exploiting the sparsity of the compact support significantly reduces training time towards enabling real-time viability.

ABSTRACT. We present a tactile inspection approach for legged robots which allows for evaluating concrete deterioration in medium to large-sized sewers. The legged robot, to this end, executes a well-defined scratching motion with one limb on the sewer floor. Inertial and Force/Torque sensors implemented in specially designed feet capture the resulting vibration. The data is used with concrete deterioration estimates from professional sewer inspectors to train a support vector machine. We achieved deterioration level estimates within three classes of over 92% accuracy. The dataset, which we recorded during a field test campaign in the sewers of Zurich is made publicly available.

ABSTRACT. Offshore HVDC converter stations for transportation of offshore wind energy to the coast need regular human inspection. Automated surveillance of such a platform by a mobile robot has high potential of improving the speed and quality of decision making while reducing operating expenses and risk of unmanned operation of the platforms. However, the challenging environment of such platforms has prevented operators from making use of mobile robots to this date. Recent progress in legged robotics resulted in systems that are becoming feasible for such tasks nowadays. For this reason, the quadrupedal robot ANYmal was tested on a platform in the North Sea for automated inspection. This paper presents the results of the field tests and discusses the challenges of industrial inspection of offshore sites.

ABSTRACT. In Japan, expectations for the automation of construction machines are increasing to solve labor shortage in the construction industry. In this research, a robotization method by retrofitting a robot to conventional construction machines is introduced to lower the introduction barrier for regional construction companies. The target machine is a six-wheeled dump truck. With a retrofitted internal sensor unit and derived kinematics of six-wheeled articulated dump truck, a conventional GNSS-based path tracking method was implemented on it. In addition, to ensure safety during operation, an emergency stop function was installed on the dump truck with three dimensional Light Detection and Ranging (3D-LiDAR). Initial experiments of forward and backward path tracking with actual dump truck confirmed the validity of the method, and the maximum tracking error was 1 m. Further, in an emergency stop experiment, the dump truck detected the obstacle and stopped immediately after obstacle detection within the emergency-stop region, i.e., 25 m x 3 m in front of the dump truck. Based on the initial experiments, the authors concluded that even the retrofitted conventional dump truck could perform basic functions for autonomous driving, such as path tracking and emergency stop.

ABSTRACT. Dry stacking with found, minimally processed rocks is a useful capability when it comes to autonomous construction. However, it is a difficult planning problem since both the state and action space are continuous, and structural stability is strongly affected by complex friction and contact constraints. We propose an algorithmic approach for autonomous construction from a collection of irregularly shaped objects. The structure planning is calculated in simulation by first considering geometric and physical constraints to find a small set of feasible actions and then refined by using a hierarchical filter based on heuristics. These plans are then executed open-loop with a robotic arm equipped with a wrist RGB-D camera. Experimental results show that the proposed planning algorithm can significantly improve the state of the art robotics dry-stacking techniques.