IROS 2019 论文列表

Iterative trajectory optimization techniques for non-linear dynamical systems are among the most powerful and sample-efficient methods of model-based reinforcement learning and approximate optimal control. By leveraging time-variant local linear-quadratic approximations of system dynamics and reward, such methods can find both a target-optimal trajectory and time-variant optimal feedback controllers. However, the local linear-quadratic assumptions are a major source of optimization bias that leads to catastrophic greedy updates, raising the issue of proper regularization. Moreover, the approximate models' disregard for any physical state-action limits of the system causes further aggravation of the problem, as the optimization moves towards unreachable areas of the state-action space. In this paper, we address the issue of constrained systems in the scenario of online-fitted stochastic linear dynamics. We propose modeling state and action physical limits as probabilistic chance constraints linear in both state and action and introduce a new trajectory optimization technique that integrates these probabilistic constraints by optimizing a relaxed quadratic program. Our empirical evaluations show a significant improvement in learning robustness, which enables our approach to perform more effective updates and avoid premature convergence observed in state-of-the-art algorithms.

This paper addresses the problem of cooperative transportation using multiple omnidirectional Automated Guided Vehicles (AGVs). To enhance flexibility and application potentials, mecanum-wheeled platform is considered for the proposed multi-AGV system while the cooperative transportation is executed without physical link/gripper to fix the object on the AGVs. Therefore, the position and number of AGV is adjustable depending on the size and weight of the transported object. Analysis of force distribution to each AGV during cooperative transportation is presented. Furthermore, the gradient projection method is exploited to regulate internal force according to the operational capability of each AGV. Moreover, an adaptive sliding mode controller is designed for AGV to cope with dynamic uncertainty during cooperative transportation. Stability of the proposed controller is proven by using Lyapunov Theorem. Finally, numerical simulation is presented to demonstrate the performance of the proposed control system.

Motivated by the recursive Newton-Euler formulation, we propose a novel cascaded Gaussian process learning framework for the inverse dynamics of robot manipulators. This approach leads to a significant dimensionality reduction which in turn results in better learning and data efficiency. We explore two formulations for the cascading: the inward and outward, both along the manipulator chain topology. The learned modeling is tested in conjunction with the classical inverse dynamics model (semi-parametric) and on its own (non-parametric) in the context of feed-forward control of the arm. Experimental results are obtained with Jaco 2 six-DOF and SARCOS seven-DOF manipulators for randomly defined sinusoidal motions of the joints in order to evaluate the performance of cascading against the standard GP learning. In addition, experiments are conducted using Jaco 2 on a task emulating a pouring maneuver. Results indicate a consistent improvement in learning speed with the inward cascaded GP model and an overall improvement in data efficiency and generalization.

Soft robots are a burgeoning archetype in robotics due to their ability to perform intricate movements easily and seamlessly. They serve as an ideal concept for realistically emulating animal movements. However, the majority of soft robots today are unable to vary their motions due to the coupled interaction of the nature of their kinematics and structural composition. We therefore created Carpie, a robotic caterpillar adapted from a modular, pneumatic actuator. Carpie is designed to perform a variety of caterpillar movements by tuning its mechanical structure through physical reconfiguration. The robot has a completely soft body that enables a variety of movements and contortions into different shapes, making Carpie is perhaps the perfect example to showcase a soft robot's aptness for animal mimicry. We analyzed the robot's control aspect. Its capabilities were measured by performing gait velocity studies on three different configurations. Each configuration was executed by applying various modules on Carpie's structure without fully refabricating the entire assembly. We also analyzed the accompanying change in the robot's maximum height during gait. Finally, we demonstrate how physical reconfiguration was able to radically alter Carpie's movement, allowing it to perform a turning motion.

Warehouse management system assigns a preferred completion time for every order based on the customer profile and the good(s) that is/are ordered. Even though employing multi-robot systems to manage goods movement bring operational efficiency in a warehouse, it is difficult to meet these soft deadlines of tasks in the peak hours/seasons. This can impact the respective businesses significantly as the lateness of task completion incurs a direct/indirect penalty. In this work, we develop an online task scheduling algorithm for such a multi-robot system, called Online Minimum Penalty Scheduling (OMPS). Though there exists a large number of multi-robot task scheduling algorithms, they are not suitable (or less efficient) for a system where each task has a soft deadline and accumulates penalty if it is executed beyond its deadline. Moreover, the lack of knowledge of future tasks (online scheduling) makes task allocation a much more difficult job. OMPS provides a robust, scalable, and near-optimal online task schedule. By comparing with the state-of-the-art algorithm, we show that OMPS attracts up to 78% less penalty when a significant number of tasks are bound to miss the deadline. Additionally, it achieves a competitive ratio of up to 1 when compared with a state-of-the-art offline task scheduling algorithm.

This paper presents a robot behavioral model designed to support children during self-study. In particular, we want to investigate how a robot could increase the time children keep concentration. The behavioral model was developed by observing children during self-study and by collecting information from experienced tutors through interviews. After observing the children, we decided to consider three states corresponding to different levels of concentration. The child can be smoothly performing the task (“learning” state), encountering some difficulties (“stuck” state) or distracted (“distracted” state). The behavioral model was designed to increase the time spent concentrating on the task by implementing adequate behaviors for each of these three states. These behaviors were designed using the advices collected during the interview survey of the experienced tutors. A self-study system based on the proposed behavior model was implemented. In this system, a small robot sits on the table and encourages the child during self-study. An operator is in charge of determining the state of the child (Wizard of Oz) and the behavioral model triggers the appropriate behaviors for the different states. To demonstrate the effectiveness of the proposed behavioral model, a user study was conducted: 22 children were asked to solve problems alone and to solve problems with the robot. The children spent significantly (p = 0.024) more time in the “learning” state when studying with the robot.

Knowing the geometry of a space is desirable for many applications, e.g. sound source localization, sound field reproduction or auralization. In circumstances where only acoustic signals can be obtained, estimating the geometry of a room is a challenging proposition. Existing methods have been proposed to reconstruct a room from the room impulse responses (RIRs). However, the sound source and microphones must be deployed in a feasible region of the room for it to work, which is impractical when the room is unknown. This work propose to employ a robot equipped with a sound source and four acoustic sensors, to follow a proposed path planning strategy to moves around the room to collect first image sources for room geometry estimation. The strategy can effectively drives the robot from a random initial location through the room so that the room geometry is guaranteed to be revealed. Effectiveness of the proposed approach is extensively validated in a synthetic environment, where the results obtained are highly promising.

Aerial filming is constantly gaining importance due to the recent advances in drone technology. It invites many intriguing, unsolved problems at the intersection of aesthetical and scientific challenges. In this work, we propose a deep reinforcement learning agent which supervises motion planning of a filming drone by making desirable shot mode selections based on aesthetical values of video shots. Unlike most of the current state-of-the-art approaches that require explicit guidance by a human expert, our drone learns how to make favorable viewpoint selections by experience. We propose a learning scheme that exploits aesthetical features of retrospective shots in order to extract a desirable policy for better prospective shots. We train our agent in realistic AirSim simulations using both a hand-crafted reward function as well as reward from direct human input. We then deploy the same agent on a real DJI M210 drone in order to test the generalization capability of our approach to real world conditions. To evaluate the success of our approach in the end, we conduct a comprehensive user study in which participants rate the shot quality of our methods. Videos of the system in action can be seen at https://youtu.be/qmVw6mfyEmw.

The aim of our study is to develop a method by which a social robot can greet passersby and get their attention without causing them to suffer discomfort. Social robots now function in a number of customer service roles, such as receptionists, guides, and exhibitors. However, sudden greetings from a robot can startle passersby. Therefore, we developed a method that allows social robots to adapt their mannerisms situationally based the results of related work. Our proposed method, user-centered reinforcement learning, enables robots to greet passersby without causing them discomfort (p<0.01). Our field experiment in an office entrance demonstrated that our method meets this requirement.

In computer vision, estimating camera pose from correspondences between 3D geometric entities and their projections into the image is a widely investigated problem. Although most state-of-the-art methods exploit simple primitives such as points or lines, and thus require dense scene models, the emergence of very effective CNN-based object detectors in the recent years have paved the way to the use of much lighter 3D models composed solely of a few semantically relevant features. In that context, we propose a novel model-based camera pose estimation method in which the scene is modeled by a set of virtual ellipsoids. We show that 6-DoF camera pose can be determined by optimizing only the three orientation parameters, and that at least two correspondences between 3D ellipsoids and their 2D projections are necessary in practice. We validate the approach on both simulated and real environments.

In this paper, we propose a noise-aware exposure control algorithm for robust robot vision. Our method aims to capture best-exposed images, which can boost the performance of various computer vision and robotics tasks. For this purpose, we carefully design an image quality metric that captures complementary quality attributes and ensures light-weight computation. Specifically, our metric consists of a combination of image gradient, entropy, and noise metrics. The synergy of these measures allows the preservation of sharp edges and rich texture in the image while maintaining a low noise level. Using this novel metric, we propose a real-time and fully automatic exposure and gain control technique based on the Nelder-Mead method. To illustrate the effectiveness of our technique, a large set of experimental results demonstrates the higher qualitative and quantitative performance compared with conventional approaches.

This paper developed a 4-degree-of-freedom cable-driven upper limb rehabilitation robot and proposed a control algorithm of the passive training for this robot. Comparing with the conventional cable-driven rehabilitation robot, the workspace of this robot is increased by optimizing the distribution of the cable attachment points and by improving the mechanical design. The rotation structure of the upper arm module can change the distribution of the attachment points as needed, by which the cable tension planner can be satisfied in almost all cases. At the meantime, the internal/external rotation of shoulder joint can be achieved without the change of the cables configuration, which is also important for increasing the workspace and comfortability of utilization. The activities of daily living (ADLs) training can be achieved well without any manual adjustment. The related controller for passive training is designed, which includes a higher controller for trajectory tracking and a lower controller for keeping cable tension as the output of the tension planner in real-time. The passive training experiments are conducted on five healthy subjects of different body size. The results demonstrated that the passive training can be achieved well on different subjects and the cable tension controller is also working effectively.

Wheeled robots operating in the field are faced with mobility challenges such as drop-offs, slippery slopes, and cluttered environments. In these situations, it is desirable to have agile and highly mobile platforms. This paper explores the utilization of multiple rotors to enhance the mobility of ground vehicles. A strategy based on downward thrust generation is developed to increase traction and surmount otherwise non-traversable slippery inclines. Both simulation and experimental results are presented to show the efficacy of the proposed solution.

Traditional attempts for loop closure detection typically use hand-crafted features, relying on geometric and visual information only, whereas more modern approaches tend to use semantic, appearance or geometric features extracted from deep convolutional neural networks (CNNs). While these approaches are successful in many applications, they do not utilize all of the information that a monocular image provides, and many of them, particularly the deep-learning based methods, require user-chosen thresholding to actually close loops - which may impact generality in practical applications. In this work, we address these issues by extracting all three modes of information from a custom deep CNN trained specifically for the task of place recognition. Our network is built upon a combination of a semantic segmentator, Variational Autoencoder (VAE) and triplet embedding network. The network is trained to construct a global feature space to describe both the visual appearance and semantic layout of an image. Then local keypoints are extracted from maximally-activated regions of low-level convolutional feature maps, and keypoint descriptors are extracted from these feature maps in a novel way that incorporates ideas from successful hand-crafted features. These keypoints are matched globally for loop closure candidates, and then used as a final geometric check to refute false positives. As a result, the proposed loop closure detection system requires no touchy thresholding, and is highly robust to false positives - achieving better precision-recall curves than the state-of-the-art NetVLAD, and with real-time speeds.

Shape morphing is at the core of future research, which shows promise for wide applications ranging from reconfigurable electronics to soft material robots. In this paper, we present a novel buckling-induced mechanism for shape morphing using dielectric elastomer actuators (DEAs), by bonding the planar precursor structure's “feet” with the DEA. With spatially-varying electric fields applied, the inplane deformation of the DEA generates compressive loads to trigger buckling of planar precursors into desired three-dimensional configurations. To enlarge the achievable motion range at the “feet”, we develop a design optimization approach to the electrode arrangement which is concisely described by cosine functions. By numerically optimizing the cosine function coefficients, we obtain the optimal spatially-varying electrodes for various precursors with different patterns of bonding sites. The experimental results demonstrate the remarkable shape-morphing from two dimensions to three dimensional configurations. Our work paves the way to novel actuation mechanisms for shape-morphing structures, with advantages of rapid response and reversible controllability.

We present a feedback motion planning algorithm, Bounded-Error LQR-Trees, that leverages reinforcement learning theory to find a policy with a bounded amount of error. The algorithm composes locally valid linear-quadratic regulators (LQR) into a nonlinear controller, similar to how LQR-Trees constructs its policy, but minimizes the cost of the constructed policy by minimizing the Bellman Residual, which is estimated in the overlapping regions of LQR controllers. We prove a sample-based upper bound on the true Bellman Residual, and demonstrate a five-fold reduction in cost over previous methods on a simple underactuated nonlinear system.

Due to implicitly introduced periodic shifting of limited searching area, visual object tracking using correlation filters often has to confront undesired boundary effect. As boundary effect severely degrade the quality of object model, it has made it a challenging task for unmanned aerial vehicles (UAV) to perform robust and accurate object following. Traditional hand-crafted features are also not precise and robust enough to describe the object in the viewing point of UAV. In this work, a novel tracker with online enhanced background learning is specifically proposed to tackle boundary effects. Real background samples are densely extracted to learn as well as update correlation filters. Spatial penalization is introduced to offset the noise introduced by exceedingly more background information so that a more accurate appearance model can be established. Meanwhile, convolutional features are extracted to provide a more comprehensive representation of the object. In order to mitigate changes of objects' appearances, multi-frame technique is applied to learn an ideal response map and verify the generated one in each frame. Exhaustive experiments were conducted on 100 challenging UAV image sequences and the proposed tracker has achieved state-of-the-art performance.

This study proposes a biped-robot pelvis-kinematics estimator based on the touch-point updating method. Because the pelvis frame is used as the base coordinate for the control, the kinematics of it with respect to the global frame should be precisely estimated. To this end, it was necessary to know where the robot made contact with the ground. The touch-point concept was introduced as the temporal contact-point, which was instantly the robot's rotation center. By updating this point, the biped's global pelvis-kinematics could be estimated. The proposed estimator was implemented into the actual robot, and its superiority was verified through ground-truth data.

Sampling-based approaches are often favored in robotics for high-dimensional motion planning for their fast exploration of the search space. However, at best they offer asymptotic guarantees on solution quality due to their inherent stochasticity. While planning, the majority of effort is often spent near the start and goal configurations with a large amount of free space in between. Bidirectional approaches such as RRT-Connect exploit this fact by greedily extending and connecting search frontiers that simultaneously propagate from the start and goal configurations of a planning problem. In this work, we use such an extend operator for bidirectional heuristic search-based planners, which typically struggle with high-dimensionality. In doing so, we address the difficulty that these bidirectional planners face with connecting frontiers of both search efforts while providing suboptimality bounds on solution quality. We validate our simple approach on high-dimensional manipulation tasks, demonstrating significantly reduced search effort when compared against other popular bidirectional algorithms, both search-based and sampling. Our algorithm maintains theoretical guarantees on suboptimality and completeness for a given resolution. In addition, the solutions found by our planner are of higher quality compared to those found by the other baseline algorithms.