IROS 2018 论文列表

This paper introduces a family of iterative algorithms for unconstrained nonlinear optimal control. We generalize the well-known iLQR algorithm to different multiple shooting variants, combining advantages like straightforward initialization and a closed-loop forward integration. All algorithms have similar computational complexity, i.e. linear complexity in the time horizon, and can be derived in the same computational framework. We compare the full-step variants of our algorithms and present several simulation examples, including a high-dimensional underactuated robot subject to contact switches. Simulation results show that our multiple shooting algorithms can achieve faster convergence, better local contraction rates and much shorter runtimes than classical iLQR, which makes them a superior choice for nonlinear model predictive control applications.

This paper proposes the implementation of an impedance controller on the ankle level of COMAN+, a robot with parallel kinematics ankles actuated by a dual four-bar mechanism. The main contribution of the work is a realization of said control scheme that grants a less abrupt and safer robot response in case of system failures, that would cause the local joint torque controllers to lose their torque reference inputs. In particular, we propose a semi-centralized impedance control implementation which eliminates the instability of the pure joint torque control schemes used in the classical fully centralized methods when torque reference interruptions occur. Finally, we present experimental results, proving the effectiveness of our method and demonstrating how it ensures a safer behaviour compared to a fully centralized impedance control implementation when the communication to the ankle joints is interrupted. This paper is a follow-up work of [1], which presented and analyzed the parallel kinematics ankles.

In this work we consider a swarm of agents shaped as bars with a certain orientation in the state space. Members of the swarm have to reach an aggregate state, while guaranteeing the collision avoidance and possibly achieving an angular consensus. By relying on a segment-to-segment distance definition, we propose a control law, which guides the agents towards this goal. A theoretical analysis of the proposed control scheme along with simulations and experimental results is provided. The proposed framework can be used to model several application scenarios ranging from collaborative transportation to precision farming, where each agent may represent either a large robot or a group of robots intent to carry bar-like shaped loads. Representative examples include: a fleet of robot-teams performing a collaborative object transportation task in an automated logistic setting, or a fleet of autonomous tractors each carrying a large atomizer to spray chemical products for pest and disease control in a precision farming setting.

This study presents a differential elastic joint for use in multi-linked pipeline inspection robots. Active joints to stretch against the pipe wall are essential for adapting robots to use in vertical pipes and slippery inner surfaces where a large traction force is required. Series elastic actuators with a high reduction system have typically been used to sense force/torque in such applications. However, compactness, power, and bi-directional series elasticity are required to conduct in-pipe inspections. In this study, we propose an active joint using a differential elastic actuator with a rubber spring for decreasing the size and increasing the stiffness of the joint. After describing the configuration of the differential elastic actuator that is suitable for our robot and the design theory of the rubber spring cross-section, we conducted experiments to verify its torque property.

Marine growths that flourish on the surfaces of underwater structures, such as bridge pylons, make the inspection and maintenance of these structures challenging. A robotic solution, using an Intervention Autonomous Underwater Vehicle (I-AUV), is developed for removing marine growth. This paper presents a Deformable Spiral Coverage Path Planning (DSCPP) algorithm for marine growth removal. DSCPP generates smooth paths to prevent damage to the surfaces of the structures and to avoid frequent or aggressive decelerations and accelerations due to sharp turns. DSCPP generates a spiral path within a circle and analytically maps the path to a minimum bounding rectangle which encompasses an area of a surface with marine growth. It aims to achieve a spiral path with minimal length while preventing missed areas of coverage. Several case studies are presented to validate the algorithm. Comparison results show that DSCPP outperforms the popular boustrophedon-based coverage approach when considering the requirements for the application under consideration.

Deep learning techniques for motion control have recently been qualitatively improved, since the successful application of Deep Q- Learning to the continuous action domain in Atari-like games. Based on these ideas, Deep Deterministic Policy Gradients (DDPG) algorithm was able to provide impressive results in continuous state and action domains, which are closely linked to most of the robotics-related tasks. In this paper, a vision-based autonomous multirotor landing maneuver on top of a moving platform is presented. The behaviour has been completely learned in simulation without prior human knowledge and by means of deep reinforcement learning techniques. Since the multirotor is controlled in attitude, no high level state estimation is required. The complete behaviour has been trained with continuous action and state spaces, and has provided proper results (landing at a maximum velocity of 2 m/s), Furthermore, it has been validated in a wide variety of conditions, for both simulated and real-flight scenarios, using a low-cost, lightweight and out-of-the-box consumer multirotor.

A common approach to the generation of walking patterns for humanoid robots consists in adopting a layered control architecture. This paper proposes an architecture composed of three nested control loops. The outer loop exploits a robot kinematic model to plan the footstep positions. In the mid layer, a predictive controller generates a Center of Mass trajectory according to the well-known table-cart model. Through a whole-body inverse kinematics algorithm, we can define joint references for position controlled walking. The outcomes of these two loops are then interpreted as inputs of a stack-of-task QP-based torque controller, which represents the inner loop of the presented control architecture. This resulting architecture allows the robot to walk also in torque control, guaranteeing higher level of compliance. Real world experiments have been carried on the humanoid robot iCub.

Autonomous robotic assisted surgery (RAS) systems aim to reduce human errors and improve patient outcomes leveraging robotic accuracy and repeatability during surgical procedures. However, full automation of RAS in complex surgical environments is still not feasible and collaboration with the surgeon is required for safe and effective use. In this work, we utilize our Smart Tissue Autonomous Robot (STAR) to develop and evaluate a shared control strategy for the collaboration of the robot with a human operator in surgical scenarios. We consider 2D pattern cutting tasks with partial blood occlusion of the cutting pattern using a robotic electrocautery tool. For this surgical task and RAS system, we i) develop a confidence-based shared control strategy, ii) assess the pattern tracking performances of manual and autonomous controls and identify the confidence models for human and robot as well as a confidence-based control allocation function, and iii) experimentally evaluate the accuracy of our proposed shared control strategy. In our experiments on porcine fat samples, by combining the best elements of autonomous robot controller with complementary skills of a human operator, our proposed control strategy improved the cutting accuracy by 6.4%, while reducing the operator work time to 44% compared to a pure manual control.

In Minimally Invasive Robotic Surgery (MIRS) a robot is interposed between the surgeon and the surgical site to increase the precision, dexterity, and to reduce surgeon's effort and cognitive load with respect to the standard laparoscopic interventions. However, the modern robotic systems for MIRS are still based on the traditional telemanipulation paradigm, e.g. the robot behaviour is fully under surgeon's control, and no autonomy or assistance is implemented. In this work, supervised and shared controllers have been developed in a vision-free, human-in-the-Ioop, control framework to help surgeon during a surgical suturing procedure. Experiments conducted on the da Vinci Research Kit robot proves the effectiveness of the method indicating also the guidelines for improving results.

In this paper, a study on nonlinear state estimation methods for Hybrid Unmanned Aerial Underwater Vehicles (HUAUVs) is presented. Based on a detailed dynamic model simulation, we analyse and elect the best nonlinear algorithm among those presented in the state-of-the-art literature addressing local derivative-free nonlinear Kalman Filters (KFs): the Unscented Kalman Filter (UKF), the Cubature Kalman Filter (CKF) and the Transformed Unscented Kalman Filter (TUKF). Here, these three nonlinear probabilistic estimators were compared in terms of the Root Mean Square Error (RMSE) and the average execution time over Monte Carlo simulations. We simulated real-world conditions for our in-production HUAUV prototype using Inertial Measurement Unit (IMU) data and state augmentation for sensor data filtering and trajectory estimation. We have concluded that the CKF proved to be the most interesting KF to low-cost on-board applications for high dimensional state spaces.

In this paper, we analyze the effects of contact models on contact-implicit trajectory optimization for manipulation. We consider three different approaches: (1)a contact model that is based on complementarity constraints, (2)a smooth contact model, and our proposed method (3) a variable smooth contact model. We compare these models in simulation in terms of physical accuracy, quality of motions, and computation time. In each case, the optimization process is initialized by setting all torque variables to zero, namely, without a meaningful initial guess. For simulations, we consider a pushing task with varying complexity for a 7 degrees-of-freedom robot arm. Our results demonstrate that the optimization based on the proposed variable smooth contact model provides a good trade-off between the physical fidelity and quality of motions at the cost of increased computation time.

In this paper, we study a compact wheeled robot that can jump while rolling. Some robots are capable of jumping or rolling, but as far as we know, those robots are not focused on jumping while rolling. We know that animals jump while running to escape from predators. Robots can move quickly by jumping while rolling. We consider a model of a robot jumping while rolling and evaluate the proposed robot. The proposed robot has two wheels and a jumping mechanism based on snap-through buckling of an elastic strip. The proposed robot can jump about 5.9 cm high and 22 cm wide on average while maintaining a traveling speed of about 1.2 m/s. The proposed robot can change the jumping angle without greatly decreasing the impulse for the moving speed.

In this paper, we present a visual simultaneous localization and mapping (SLAM) system for humanoid robots. We introduce a new binary descriptor called DLab that exploits the combined information of color, depth, and intensity to achieve robustness with respect to uniqueness, reproducibility, and stability. We use DLab within ORB-SLAM, where we replaced the place recognition module with a modification of FAB-MAP that works with newly built codebooks using our binary descriptor. In experiments carried out in simulation and with a real Nao humanoid equipped with an RGB-D camera, we show that DLab has a superior performance in comparison to other descriptors. The application to feature tracking and place recognition reveal that the new descriptor is able to reliably track features even in sequences with seriously blurred images and that it has a higher percentage of correctly identified similar images. As a result, our new visual SLAM system has a lower absolute trajectory error in comparison to ORB-SLAM and is able to accurately track the robot's trajectory.

With a light and slender body, a cable-driven redundant spatial manipulator (CRSM) has flexible manipulability and high maneuverability in confined environment. However, compared with revolute rigid manipulators, such type of manipulators generally has low stiffness and weak load capacity. In this paper, we propose a new mechanism design to improve the stiffness and load capacity without sacrificing the manipulator dexterity and the end-effector accuracy. The manipulator is composed of 3 active-passive-linkage segments and 1 active tool end-effector. Each active-passive segment has 2 degrees of freedom (DOFs) driven by three evenly distributed cables. Pretension mechanism and linkage cables are designed to keep strict equal angles of adjacent joints. A separable control box, which contains all the motors and cable transmission mechanisms is also designed with a quick release-and-lock mechanism. Therefore, the robotic arm can be easily removed and installed. Based on the equal angle characteristic, kinematic equations of manipulator are established with Denavit-Hartenberg (D-H) method and the Jacobian matrix is also simplified. Further analysis of the workspace supplies the guidance for the task design and motion planning. Finally, a prototype system is developed to perform the stiffness and load capacity experiments. Experimental results show that the developed CRSM has relatively high stiffness and load capacity.

Navigation through tubular environment is fundamental in tasks such as pipeline inspection, gastrointestinal tract inspection, etc. Conventional pipeline inspection robots are mostly made by rigid materials and could not well adapt to the large size variation of the environment. Soft robots provide an additional solution for inspection of pipelines, especially with significant size variation. In this work, we present a soft robot for pipeline inspection, which consists of an earthworm-like soft robot and a Central Pattern Generator (CPG)-based control system. An analytical model is developed to predict the maximum pipe diameter that the robot could adapt to. For the current prototype, the robot could adapt to size change of three times. Experimental results show that this robot could navigate through pipelines with sharp turnings and with large diameter change.

Most approaches for optimisation of neural networks are based on variants of back-propagation. This requires the network to be time invariant and differentiable; neural networks with dynamics are thus generally outside the scope of these methods. Biological neural circuits are highly dynamic yet clearly able to support learning. We propose a reinforcement learning approach inspired by the mechanisms and dynamics of biological synapses. The network weights undergo spontaneous fluctuations, and a reward signal modulates the centre and amplitude of fluctuations to converge to a desired network behaviour. We test the new learning rule on a 2D bipedal walking simulation, using a control system that combines a recurrent neural network, a bio-inspired central pattern generator layer and proportional-integral control, and demonstrate the first successful solution to this benchmark task.

This paper presents a framework for engaging a target while approaching it from a long distance, using observation from sensors on-board a mobile robot. The proposed framework consists of two multi-stage Bayesian approaches to reliably detect and accurately engage with the target under uncertainties. The multi-stage localization approach localizes the robot and the target in a global coordinate frame. Their locations are estimated sequentially when the robot is at a long distance from the target, whereas they are localized simultaneously when the target is in the close vicinity. In the multi-stage target observation approach, a level of confidence and the associated probability of detection of the sensor are defined to make the target detectable in maximal occasions. This allows the extended Kalman filter to be implemented for the target engagement. The proposed framework was implemented on an unmanned ground vehicle equipped with multiple sensors. Results show the effectiveness of the proposed framework in solving real-world problems.

This paper presents a 2D B-spline mapping framework for representing unstructured environments in a compact manner. While occupancy-grid and landmark-based maps have been successfully employed by the robotics community in indoor scenarios, outdoor long-term autonomous operations require a more compact representation of the environment. This work tackles this problem by interpolating the data of a high frequency sensor using B-spline curves. Compared to lines and circles, splines are more powerful in the sense that they allow for the description of more complex shapes in the scene. In this work, spline curves are continuously tracked and aligned across multiple sensor readings using lightweight methods, making the proposed framework suitable for robot navigation in outdoor missions. In particular, a Simultaneous Localization and Mapping (SLAM) algorithm specifically tailored for B-spline maps is presented here. The efficacy of the proposed framework is demonstrated by Software-in-the-Loop (SiL) simulations in different scenarios.

Till date, various seven-degrees-of-freedom (7-DOF)robot arms have been developed globally. In general, the robot arms utilized in factories are required to possess speed, power, and accuracy. They are isolated from humans using a fence to ensure human safety. However, a home robot should work near humans at an appropriate speed without a fence. One approach to meeting this requirement involves the installation of various sensors on the robot to control and stop the robot safely even if a collision has occurred. However, a robot system is complicated and expensive. In order to give inherent safety to the robot, we should lighten the robot arm. In this study, a 7-DOF lightweight arm with a wide wrist-motion range is introduced. The weight of movable parts is approximately 2.87 kg. To achieve such properties, we propose the use of three mechanisms: a shoulder mechanism with hollow cylinders, a wrist mechanism with a wide workspace, and an integrated wrist and elbow mechanisms of high power. To verify its feasibility, a prototype of the robot is designed and developed. The experimental results demonstrate its powerful performance and wide workspace of the wrist.