ICRA 2018 Papers

This paper presents a novel cable-driven gripper with perception capabilities for autonomous harvesting of strawberries. Experiments show that the gripper allows for more accurate and faster picking of strawberries compared to existing systems. The gripper consists of four functional parts for sensing, picking, transmission, and storing. It has six fingers that open to form a closed space to swallow a target strawberry and push other surrounding berries away from the target. Equipped with three IR sensors, the gripper controls a manipulator arm to correct for positional error, and can thus pick strawberries that are not exactly localized by the vision algorithm, improving the robustness. Experiments show that the gripper is gentle on the berries as it merely cuts the stem and there is no physical interaction with the berries during the cutting process. We show that the gripper has close-to-perfect successful picking rate when addressing isolated strawberries. By including internal perception, we get high positional error tolerance, and avoid using slow, high-level closed-loop control. Moreover, the gripper can store several berries, which reduces the overall travel distance for the manipulator, and decreases the time needed to pick a single strawberry substantially. The experiments show that the gripper design decreased picking execution time noticeably compared to results found in literature.

This paper describes skating locomotion for a quadruped robot with dexterous limbs. The emphasis is on design of omnidirectional motion primitives and quantification of resulting speed and accuracy when traversing different types of smooth but potentially non-flat terrain. In particular, we study trade-offs between using four-wheeled versus three-wheeled skating maneuvers. In four-wheeled skating, motions have the benefit of symmetry, so that errors due to wheel slip should theoretically cancel out on average. Three-wheeled skating, by contrast, introduces significantly more asymmetry in configuration and contact force distribution over time; however, it has the advantage of guaranteeing continuous ground contact for all skates when terrain has bumps or other curvature. We present simulation results quantifying errors for each approach, for various terrains. Our results allow us to tune motions to reduce biases and variability in motion, which are primarily due to accelerations as locomotion begins.

Tendon-driven hand orthoses have advantages over exoskeletons with respect to wearability and safety because of their low-profile design and ability to fit a range of patients without requiring custom joint alignment. However, no existing study on a wearable tendon-driven hand orthosis for stroke patients presents evidence that such devices can overcome spasticity given repeated use and fatigue, or discusses transmission efficiency. In this study, we propose two designs that provide effective force transmission by increasing moment arms around finger joints. We evaluate the designs with geometric models and experiment using a 3D-printed artificial finger to find force and joint angle characteristics of the suggested structures. We also perform clinical tests with stroke patients to demonstrate the feasibility of the designs. The testing supports the hypothesis that the proposed designs efficiently elicit extension of the digits in patients with spasticity as compared to existing baselines.

This paper describes the design of a powered knee-and-ankle transfemoral prosthetic leg, which implements high torque density actuators with low-reduction transmissions. The low reduction of the transmission coupled with a high-torque and low-speed motor creates an actuator with low mechanical impedance and high backdrivability. This style of actuation presents several possible benefits over modern actuation styles implemented in emerging robotic prosthetic legs. Such benefits include free-swinging knee motion, compliance with the ground, negligible unmodeled actuator dynamics, and greater potential for power regeneration. Benchtop validation experiments were conducted to verify some of these benefits. Backdrive and free-swinging knee tests confirm that both joints can be backdriven by small torques (~3 Nm). Bandwidth tests reveal that the actuator is capable of achieving frequencies required for walking and running. Lastly, open-loop impedance control tests prove that the intrinsic impedance and unmodeled dynamics of the actuator are sufficiently small to control joint impedance without torque feedback.

This paper presents the design of a wearable robotic forearm for close-range human-robot collaboration. The robot's function is to serve as a lightweight supernumerary third arm for shared workspace activities. We present a functional prototype resulting from an iterative design process including several user studies. An analysis of the robot's kinematics shows an increase in reachable workspace by 246 % compared to the natural human reach. The robot's degrees of freedom and range of motion support a variety of usage scenarios with the robot as a collaborative tool, including self-handovers, fetching objects while the human's hands are occupied, assisting human-human collaboration, and stabilizing an object. We analyze the bio-mechanical loads for these scenarios and find that the design is able to operate within human ergonomic wear limits. We then report on a pilot human-robot interaction study that indicates robot autonomy is more task-time efficient and preferred by users when compared to direct voice-control. These results suggest that the design presented here is a promising configuration for a lightweight wearable robotic augmentation device, and can serve as a basis for further research into human-wearable collaboration.

Underwater gliders are special class of autonomous underwater vehicles (AUVs) proven to be power efficient with better range and endurance compared to the conventional underwater robots. Most of the existing underwater gliders use `change of mass' based variable buoyancy (VB) method in which the overall system architecture and construction are complex. A novel underwater glider RoBuoy based on the `change of volume' concept of variable buoyancy method is presented here. RoBuoy uses actuated metallic bellows to change the volume which makes the system simple and modular in construction without any compromise in the performance. It uses minimal number of parts compared to the existing gliders which reduces the overall complexity of the system. Also, most of the conventional gliders use the external fluid for its working which may result in corrosion or fouling of parts and requires frequent maintenance. In the proposed glider, all the vital parts required for its working, apart from the sensing payloads are enclosed inside the hull, thereby increasing the durability. In this paper, a detailed design of RoBuoy is discussed with its possible modes of operation. An integrated mathematical model considering the individual dynamics of the actuator, hull/fuselage, and the wings has been developed and the open loop performance of the glider is studied at different input conditions. An experimental prototype has been designed and fabricated based on optimized dimensions, with the required mechatronic system. Experiments have been conducted and the results prove the feasibility of the concept.

In this paper, we describe the design and analysis of a fixed-wing unmanned aerial-aquatic vehicle. Inspired by prior work in aerobatic post-stall maneuvers for fixed-wing vehicles [1], we explore the feasibility of executing a water-to-air transition with a fixed-wing vehicle using almost entirely commercial off-the-shelf components (excluding the fuselage). To do this, we first propose a conceptual design based on observations about the dominant forces and dimensionless analysis. We then further refine this concept by building a design tool based on simplified models to explore the design space. To verify the results of the design tool, we use a higher fidelity model along with a direct hybrid trajectory optimization approach to show via numerical simulation that the water-to-air transition is feasible. Finally, we successfully test our design experimentally by hand-piloting a prototype vehicle through the water-to-air transition and discuss our approach for replacing the human-pilot with closed-loop control.

Current multi-backbone continuum robots are limited to a constant cross-sectional diameter. This paper proposes a design alternative that overcomes this limitation. The ability to change the diameter of a continuum robot expands the repertoire of kinematic redundancy and enables kinematic parameter adaptation to optimize performance. A continuum robot design based on the angulated scissor mechanism is presented along with its position analysis. An exploration of admissible design parameter values for a given continuum robot segment with a desired maximum curvature while maintaining an open bore along its center is also carried out. A design presenting how this mechanism can be incorporated into a continuum robot is shown and a strategy for minimizing joint forces and avoiding joint limits is formulated as a gradient descent redundancy resolution problem in a simulation case study. The simulation results show that varying the diameter can significantly reduce joint forces while preserving the workspace and avoiding joint limits. This work is a first step towards continuum robots with situational awareness that will use their sensing capabilities to adapt their structure in order to optimize task execution performance.

This paper addresses the problem of multi-robot path coordination, considering specific features that arise in applications such as automatic aircraft taxiing or driver-less cars coordination. The first feature is departure events: when robots arrive at their destinations (e.g. the runway for takeoff), they can be removed from the coordination diagram. The second feature is the “no-backward-movement” constraint: the robots can only move forward on their assigned paths. These features can interact to give rise to complex conflict situations, which existing planners are unable to solve in practical time. We propose a set of algorithms to efficiently account for these features and validate these algorithms on a realistic model of Charles de Gaulle airport.

In this work, we develop a novel dense planar-inertial SLAM (DPI-SLAM) system to reconstruct dense 3D models of large indoor environments using a hand-held RGB-D sensor and an inertial measurement unit (IMU). The preinte-grated IMU measurements are loosely-coupled with the dense visual odometry (VO) estimation and tightly-coupled with the planar measurements in a full SLAM framework. The poses, velocities, and IMU biases are optimized together with the planar landmarks in a global factor graph using incremental smoothing and mapping with the Bayes Tree (iSAM2). With odometry estimation using both RGB-D and IMU data, our system can keep track of the poses of the sensors even without sufficient planes or visual information (e.g. textureless walls) temporarily. Modeling planes and IMU states in the fully probabilistic global optimization reduces the drift that distorts the reconstruction results of other SLAM algorithms. Moreover, structural constraints between nearby planes (e.g. right angles) are added into the DPI-SLAM system, which further recovers the drift and distortion. We test our DPI-SLAM on large indoor datasets and demonstrate its state-of-the-art performance as the first planar-inertial SLAM system.

Place recognition is a key element of mobile robotics. It can assist with the “wake-up” and “kidnapped robot” problems, where the robot position needs to be estimated without prior information. Among the different sensors that can be used for the task (e.g., camera, GPS, LiDAR), LiDAR has the advantage of operating in the dark and in GPS-denied areas. We propose a new method that uses solely the LiDAR data and that can be performed without robot motion. In contrast to other methods, our system leverages intensity information (as opposed to only range information) which is encoded into a novel descriptor of LiDAR intensities as a group of histograms, named DELIGHT. The descriptor encodes the distributed histograms of intensity of the surroundings which are compared using chi-squared tests. Our pipeline is a two-stage solution consisting of an intensity-based prior estimation and a geometry-based verification. For a map of 220k square meters, the method achieves localisation in around 3s with a success rate of 97%, illustrating the applicability of the method in real environments.

We propose a novel approach to detect vanishing points in images using a convolutional neural network (CNN) trained on a newly collected Google street-view image dataset. By utilizing the camera parameters and road direction data from Google street view, we collected a total of 1,053,425 images with inferred ground-truth vanishing points, along 23 worldwide routes totaling 125,165 kilometers. We then formulate vanishing point detection as a CNN classification problem using an output layer with 225 discrete possible vanishing point locations. Experimental results show that our deep vanishing point system outperforms the state-of-the-art algorithmic vanishing point detector. We achieved 99% accuracy in recovering the horizon line and 92% in locating the vanishing point within a ±5-degree range.

In this paper, we present a new approach for efficient regression-based object tracking. Our approach is closely related to the Generic Object Tracking Using Regression Networks (GOTURN) framework [1]. We make the following contributions. First, we demonstrate that there is a theoretical relationship between Siamese regression networks like GOTURN and the classical Inverse Compositional Lucas & Kanade (IC-LK) algorithm. Further, we demonstrate that unlike GOTURN, IC-LK adapts its regressor to the appearance of the current tracked frame. We argue that the lack of such property in GOTURN attributes to its poor performance on unseen objects and/or viewpoints. Second, we propose a novel framework for object tracking inspired by the IC-LK framework, which we refer to as Deep-LK. Finally, we show impressive results demonstrating that Deep-LK substantially outperforms GOTURN and demonstrate comparable tracking performance against current state-of-the-art deep trackers on high frame-rate sequences whilst being an order of magnitude (100 FPS) computationally efficient.

Navigation in pedestrian environments is critical to enabling independent mobility for the blind and visually impaired (BVI) in their daily lives. White canes have been commonly used to obtain contact feedback for following walls, curbs, or man-made trails, whereas guide dogs can assist in avoiding physical contact with obstacles or other pedestrians. However, the infrastructures of tactile trails or guide dogs are expensive to maintain. Inspired by the autonomous lane following of self-driving cars, we wished to combine the capabilities of existing navigation solutions for BVI users. We proposed an autonomous, trail-following robotic guide dog that would be robust to variances of background textures, illuminations, and interclass trail variations. A deep convolutional neural network (CNN) is trained from both the virtual and realworld environments. Our work included major contributions: 1) conducting experiments to verify that the performance of our models trained in virtual worlds was comparable to that of models trained in the real world; 2) conducting user studies with 10 blind users to verify that the proposed robotic guide dog could effectively assist them in reliably following man-made trails.

In this paper, we explore deep reinforcement learning algorithms for vision-based robotic grasping. Model-free deep reinforcement learning (RL) has been successfully applied to a range of challenging environments, but the proliferation of algorithms makes it difficult to discern which particular approach would be best suited for a rich, diverse task like grasping. To answer this question, we propose a simulated benchmark for robotic grasping that emphasizes off-policy learning and generalization to unseen objects. Off-policy learning enables utilization of grasping data over a wide variety of objects, and diversity is important to enable the method to generalize to new objects that were not seen during training. We evaluate the benchmark tasks against a variety of Q-function estimation methods, a method previously proposed for robotic grasping with deep neural network models, and a novel approach based on a combination of Monte Carlo return estimation and an off-policy correction. Our results indicate that several simple methods provide a surprisingly strong competitor to popular algorithms such as double Q-learning, and our analysis of stability sheds light on the relative tradeoffs between the algorithms1.

Exploration region selection is an essential decision making process in autonomous robot exploration task. While a majority of greedy methods are proposed to deal with this problem, few efforts are made to investigate the importance of predicting long-term planning. In this paper, we present an algorithm that utilizes deep reinforcement learning (DRL) to learn exploration knowledge over office blueprints, which enables the agent to predict a long-term visiting order for unexplored subregions. On the basis of this algorithm, we propose an exploration architecture that integrates a DRL model, a next-best-view (NBV) selection approach and a structural integrity measurement to further improve the exploration performance. At the end of this paper, we evaluate the proposed architecture against other methods on several new office maps, showing that the agent can efficiently explore uncertain regions with a shorter path and smarter behaviors.

In this paper, we investigate a predictive approach for collision risk assessment in autonomous and assisted driving. A deep predictive model is trained to anticipate imminent accidents from traditional video streams. In particular, the model learns to identify cues in RGB images that are predictive of hazardous upcoming situations. In contrast to previous work, our approach incorporates (a) temporal information during decision making, (b) multi-modal information about the environment, as well as the proprioceptive state and steering actions of the controlled vehicle, and (c) information about the uncertainty inherent to the task. To this end, we discuss Deep Predictive Models and present an implementation using a Bayesian Convolutional LSTM. Experiments in a simple simulation environment show that the approach can learn to predict impending accidents with reasonable accuracy, especially when multiple cameras are used as input sources.

Robotic manipulation in complex open-world scenarios requires both reliable physical manipulation skills and effective and generalizable perception. In this paper, we propose using an object-centric prior and a semantic feature space for the perception system of a learned policy. We devise an object-level attentional mechanism that can be used to determine relevant objects from a few trajectories or demonstrations, and then immediately incorporate those objects into a learned policy. A task-independent attention locates possible objects in the scene, and a task-specific attention identifies which objects are predictive of the trajectories. The scope of the task-specific attention is easily adjusted by showing demonstrations with distractor objects or with diverse relevant objects. Our results indicate that this approach exhibits good generalization across object instances using very few samples, and can be used to learn a variety of manipulation tasks using reinforcement learning.

We propose to use neural networks for simultaneous detection and localization of multiple sound sources in human-robot interaction. In contrast to conventional signal processing techniques, neural network-based sound source localization methods require fewer strong assumptions about the environment. Previous neural network-based methods have been focusing on localizing a single sound source, which do not extend to multiple sources in terms of detection and localization. In this paper, we thus propose a likelihood-based encoding of the network output, which naturally allows the detection of an arbitrary number of sources. In addition, we investigate the use of sub-band cross-correlation information as features for better localization in sound mixtures, as well as three different network architectures based on different motivations. Experiments on real data recorded from a robot show that our proposed methods significantly outperform the popular spatial spectrum-based approaches.