IROS 2018 Papers

With the advancements in robotic technology and the progress in human-robot interaction research, the interest in deploying mixed human-robot teams in rescue missions is increasing. Due to their complementary capabilities in terms of locomotion, visibility and reachability of areas, human-robot teams are considerably deployed in real-world settings, albeit the robotic agents in such scenarios are normally fully teleoperated. A major barrier to successful and efficient mission execution in those teams is the lack of cognitive skills in robotic systems. In this paper, we present a cognition-enabled framework and an implemented system where robotic agents are equipped with cognitive capabilities to naturally communicate with humans and autonomously perform tasks. The framework allows for natural tasking of robots, reasoning about robot behavior, capabilities and actions, and a common belief state representation for shared mission awareness of robots and human operators.

This paper explores ethical and legal implications arising from the intertwinement of cloud services, healthcare and robotics. It closes an existing gap in the literature by highlighting the distinctive ethical and legal concerns associated with the inter-dependence of the cyber- and the physical aspects of healthcare cloud robotics. The identified core concerns include uncertainties with regard to data protection requirements; distributed responsibilities for unintended harm; achievement of transparency and consent for cloud robot services especially for vulnerable robot users; secondary uses of cloud data derived from robot activities; data security; and wider social issues. The paper aims to raise awareness and stimulate reflection of the legal and ethical impacts on different stakeholders arising from the use of cloud services in healthcare robotics. We show that due to the complexity of these concerns the design and implementation of such robots in healthcare requires an interdisciplinary development and impact assessment process. In light of legal requirements and ethical responsibilities towards end-users and other stakeholders, we draw practical considerations for engineers developing cloud services for robots in healthcare.

Coiled actuators made from polymer fibers are fast becoming popular due to their low-cost and ease of fabrication. Unfortunately, reliable real-time temperature measurement has been frustrated by the small fiber diameter of typical actuators. By using coiled polymer fibers wrapped with a metal wire, we demonstrate the ability to concurrently drive a muscle by electrothermal heating, and monitor muscle temperature through the wire resistance. This simple method enables convenient overheat protection for these muscles, as well as the possibility for closed-loop temperature control. Using this platform, we demonstrate a nested controller using temperature and position feedback to improve contraction speed, and investigate the cooling rates of various configurations that increase total force output.

Accurate localization is critical for a robotic vehicle to navigate autonomously. Conventional autonomous underwater vehicles (AUV s) typically rely on an inertial navigation system (INS) aided by a Doppler velocity log (DVL) in order to reduce the rate of positional error growth of dead-reckoning to a level suitable for reliable navigation underwater. The size, cost, and power requirements of these systems result in vehicles that are prohibitively large and expensive for multi-AUV operations. In this work we present the first results of closed-loop experiments using a miniature, low-cost SandShark AUV and a custom-designed, inexpensive acoustic system first described in our previous work. Results are validated using an independent LBL system, and indicate that our approach is suitably accurate to enable the self-localization of such AUVs without the use of an expensive DVL-aided INS. Self-localization is performed by obtaining acoustic range and angle measurements from the AUV to a single acoustic beacon using a vehicle-mounted passive hydrophone receiver-array, and fusing these measurements using a particle filter. A critical aspect of our approach that allows for real-time, closed-loop operation is the close coupling of conventional phased-array beamforming and particle filtering - this implementation detail reduces the computational complexity associated with our previously described two-stage beamforming plus particle filtering process, and consequently also enables an increase in particle count and an improvement in navigational accuracy. Experimental results are provided for two cases: first, absolute navigation in the case where the beacon is fixed at a known position; and second, relative navigation with a moving beacon, a novel operating paradigm for AUVs which promises to enable multi-AUV operations while maintaining bounded navigation error.

Increasing the accessibility of autonomous robots also for inexperienced users requires user-friendly and high-level control opportunities of robotic systems. While automated planning is able to decompose a complex task into a sequence of steps which reaches an intended goal, it is difficult to formulate such a goal without knowing the internals of the planning system and the exact capabilities of the robot. This becomes even more important in dynamic environments in which manipulable objects are subject to change. In this paper, we present an adaptive control interface which allows users to specify goals based on an internal world model by incrementally building referring expressions to the objects in the world. We consider fetch-and-carry tasks and automatically deduce potential high-level goals from the world model to make them available to the user. Based on its perceptions our system can react to changes in the environment by adapting the goal formulation within the domain-independent planning system.

This paper presents new schemes to estimate 3D rotation from spherical images. Unlike existing approaches, spherical moment properties are exploited to obtain a closed form solution without iteratively mimimizing a cost function. Actually, three methods using spherical moments are proposed: two of them can be classified as dense approaches, while the third one is hybrid combining geometrical features with dense ones. Experimental results using both synthetic images and acquired images using catadioptric cameras with different scenarios show the effectiveness of our approach.

Many agricultural tasks are known to be dangerous for human operators, the environment, and human health in general. The increasing pressure both on safety and on production levels motivates the development of new methodologies and technologies. The rising of off-road mobile robots for agricultural application appears to be a promising contribution to required innovations. It both permits to limit the exposure of people to hazardous products and to achieve difficult and repetitive tasks. Nevertheless, to be fully efficient, autonomous robots have to ensure a high level of accuracy, while carrying potentially heavy tools, possibly in harsh conditions. It is especially the case of spraying, for which accuracy is a key challenge for reducing environmental impacts. The use of huge robots for spraying might seem to be a straightforward solution, by simply automating existing machines. Nevertheless, a simple automation does not reduce directly the environmental impact of human activities (soil compaction, energy, reduction of the use of chemical products). Moreover, huge machines are not necessarily an advantage when considering safety aspects (rollover risk and maneuverability). As a result, a solution based on the cooperation of at least two mobile robots, moving from either side of a vine row, is investigated in this paper thanks to Ultra Wide Band (UWB) technology.

The presented work deals with classification of garment categories including pants, shorts, shirts, T-shirts and towels. The knowledge of the garment category is crucial for its robotic manipulation. Our work focuses particularly on garments being held in a hanging state by a robotic arm. The input of our method is a set of depth maps taken from different viewpoints around the garment. The depths are fused into a single 3D point cloud. The cloud is fed into a convolutional neural network that transforms it into a single global feature vector. The network utilizes a generalized convolution operation defined over the local neighborhood of a point. It can deal with permutations of the input points. It was trained on a large dataset of common 3D objects. The extracted feature vector is classified with SVM trained on smaller datasets of garments. The proposed method was evaluated on publicly available data and compared to the original methods, achieving competitive performance and better generalization capability.

People's responses to a hypnosis intervention is diverse and unpredictable. A system that predicts user's level of susceptibility from their electroencephalography (EEG) signals can be helpful in clinical hypnotherapy sessions. In this paper, we extracted differential entropy (DE) of the recorded EEGs from two groups of subjects with high and low hypnotic susceptibility and built a support vector machine on these DE features for the classification of susceptibility trait. Moreover, we proposed a clustering-based feature refinement strategy to improve the estimation of such trait. Results showed a high classification performance in detection of subjects' level of susceptibility before and during hypnosis. Our results suggest the usefulness of this classifier in development of future Bel systems applied in the domain of therapy and healthcare.

Road networks in cities are massive and is a critical component of mobility. Fast response to defects, that can occur not only due to regular wear and tear but also because of extreme events like storms, is essential. Hence there is a need for an automated system that is quick, scalable and cost-effective for gathering information about defects. We propose a system for city-scale road audit, using some of the most recent developments in deep learning and semantic segmentation. For building and benchmarking the system, we curated a dataset which has annotations required for road defects. However, many of the labels required for road audit have high ambiguity which we overcome by proposing a label hierarchy. We also propose a multi-step deep learning model that segments the road, subdivide the road further into defects, tags the frame for each defect and finally localizes the defects on a map gathered using GPS. We analyze and evaluate the models on image tagging as well as segmentation at different levels of the label hierarchy.

In this paper, we demonstrate how behavioural structure, such as a finite state machine, can emerge in minimal robots without computation nor memory capabilities. As a case study we observe the ability of a group of non-holonomic robots to form robust, self-healing circle formations in a decentralized manner using only a limited frontal binary sensor. We present a grid-search method to find suitable parameters that promote the formation of a stable circle. We then examine how the parameters of the controllers affect the appearance of the behaviour, and provide theoretical proof for its emergence and self-healing properties. We validate the proposed model through a set of experiments with ten mobile real robots. Our results with real robots match the simulated experiments and provide insights on how a simple, computation-free behaviour can generate complex spatio-temporal dynamics.

This paper describes a novel passive exoskeleton for voluntary sitting-standing posture transition for children with lower limb impairment. The design of the exoskeleton is based on the utilization of the center of gravity transition through the user's upper body motion. The passive exoskeleton powered by gas springs allows the user to realize a natural-like posture transition by the user's voluntary upper body motion. We designed the posture transition model such that the user can realize the posture transition in a natural manner based on a minimum jerk criterion. The proposed design aims to permit toilet usage without transferring seating positions between the exoskeleton and toilet seat. Furthermore, the developed mechanism can be integrated with a regular wheelchair, which would allow users to have locomotion capability. We believe that the design can improve children's self-reliant social activities by supporting their voluntary posture transition and toilet use. In this paper, we describe the detailed design process of the exoskeleton and preliminary experiments to investigate its effectiveness through evaluation with a healthy participant.

Assistive devices have been developed for power augmentation and task-oriented assistance such as loaded walking. The effective joint dynamics of the user can be altered using a wearable system, providing assistance when a task is performed. The authors have investigated an Active/Passive Pneumatic Actuator (AP2A) for an assistive device, which has a simple structure and responds as a passive nonlinear spring with controllable stiffness. This paper introduces a novel controller for the AP2 A and validates the performance through experiments. The developed controller is found to stabilize at the desired stiffness response within 1 second, confirming the ability of the AP2 A to act as an adjustable passive nonlinear spring.

Indoor navigation has been a major focus of drone research over the last few decades. The main reason for the term “indoor” came from the fact that in outdoor environments, drones could rely on global navigation systems such as GPS for their position and velocity estimates. By focusing on unknown indoor environments, the research had to focus on solutions using onboard sensors and processing. In this article, we present an overview of the state of the art and remaining challenges in this area, with a focus on small drones.

Motivated by the potential of bimodal aerial and surface locomotion as an energy saving strategy for small flying robots, we investigate the effects of a flat overhang surface in the vicinity of a spinning propeller. We employ the classical momentum theory and the blade element method to describe the “ceiling effects” in regards to the generated thrust, power, and rotational speed of the propeller in terms of a normalized distance between the ceiling and the propeller. Validating experiments were performed on a benchtop setup, and the results are in agreement with the proposed models. The presence of a ceiling was found to reduce the power consumption by more than a factor of three for the same thrust force. Overall, our findings show promise, paving the way for the use of perching maneuvers by small rotorcraft to extend their missions.

The quasi-static catenary curve of a semi-slack tether between an essentially stationary unmanned air vehicle (UAV) and a small unmanned surface vehicle (USV) is investigated and characterized. An empirical analysis, performed over a discretized space of vertical and horizontal separations of the two vehicles, determines an optimum cable length & tension for maximizing system robustness during the vertical heave of the USV due to high seas. Operating at this optimum condition allows for equal displacements of the USV in the up and down directions, minimizing the possibility of both fouling (with the tether touching the water) and excessive downforce on the UAV (with the tether pulled taut) during dynamic heave events. Scaling the horizontal offset, tether length, and tension by the flying height collapses all empirical results into convenient curves depending only on a nondimensional relative position parameter (Δx/Δy), accurately fit by low order polynomials. This eliminates the need for a lookup table, and decreases computation time during implementation. The heave robustness analysis results in a recommended operating relative position of Δx/Δy ≈ .46. Experimental results are presented and confirm the catenary analysis for the proposed tether.

This paper presents a method for tracking multiple interacting deformable objects undergoing rigid motions, elastic deformations and contacts, using image and point cloud data provided by an RGB-D sensor. A joint registration frame-work is proposed, based on physical Finite Element Method (FEM) elastic and interaction models. It first relies on a visual segmentation of the considered objects in the RGB images. The different segmented point clouds are then processed to estimate rigid transformations with on an ICP algorithm, and to determine geometrical point-to-point correspondences with the meshes. External forces resulting from these correspondences and between the current and the rigidly transformed mesh can then be derived. It provides both non-rigid and rigid data cues. A classical collision detection and response model is also integrated, giving contact forces between the objects. The deformations of the objects are estimated by solving a dynamic system balancing these external and contact forces with the internal or regularization forces computed through the FEM elastic model. This approach has been here tested on different scenarios involving two or three interacting deformable objects of various shapes, with promising results.

In this paper, we present a method for contactless material detection with capacitive proximity sensing skins. Our new approach extends the current state-of-the-art proximity and distance sensing methods and measures the characteristic impedance spectrum of an object to obtain material properties. By this, we gain further material information besides of the near field information in a contactless and non-destructive way. The measurement method requires sensors that provide absolute distance and frequency based capacitance measurement capabilities and can be applied to similar systems. The sensor system described in this paper measures proximity with a capacitance based sensor and absolute distance based on time-of-flight (ToF)sensors. Attached on a robot, we gain information about the robot's near field environment. The information is important not only for human- machine- interaction, but also for grasping and manipulation. We focus on signal processing and evaluate our method with measurements of numerous different materials and present a solution to differentiate between them.

3D LiDARs and 2D cameras are increasingly being used alongside each other in sensor rigs for perception tasks. Before these sensors can be used to gather meaningful data, however, their extrinsics (and intrinsics) need to be accurately calibrated, as the performance of the sensor rig is extremely sensitive to these calibration parameters. A vast majority of existing calibration techniques require significant amounts of data and/or calibration targets and human effort, severely impacting their applicability in large-scale production systems. We address this gap with CalibNet: a geometrically supervised deep network capable of automatically estimating the 6-DoF rigid body transformation between a 3D LiDAR and a 2D camera in real-time. CalibNet alleviates the need for calibration targets, thereby resulting in significant savings in calibration efforts. During training, the network only takes as input a LiDAR point cloud, the corresponding monocular image, and the camera calibration matrix K. At train time, we do not impose direct supervision (i.e., we do not directly regress to the calibration parameters, for example). Instead, we train the network to predict calibration parameters that maximize the geometric and photometric consistency of the input images and point clouds. CalibNet learns to iteratively solve the underlying geometric problem and accurately predicts extrinsic calibration parameters for a wide range of mis-calibrations, without requiring retraining or domain adaptation. The project page is hosted at https://epiception.github.io/CalibNet.