IROS 2016 论文列表

This paper is aimed at developing optimal motion planning for a single autonomous surface vehicle (ASV) equipped with an on-board pollutant sensor that will maximize the sensor-related information available for source seeking. The ASV uses a nonlinear diffusion model of the pollutant source to estimate the intensity/level of the pollution at the present ASV location. The rate of detection of particles depends on the relative distance between the ASV and the source. First, we use a probabilistic map of the source location built through the sensor information for a dynamic motion planning of source seeking based on an entropy reduction formulation, where an appropriately defined Fisher information matrix (FIM) is used for entropy reduction or information gain. We derive the FIM for the set-up and investigate optimal trajectories. Next, we present an online nonlinear Monte Carlo algorithm that uses the obtained sensor information about pollutant at different vehicle locations to update a probabilistic uncertainty map of pollutant source location. As the mission unfolds the ASV motion is computed by considering a moving-horizon interval of decision, which will allow for the inclusion of new information available for optimal motion planning. The proposed motion planning approach is extended to take into account external disturbances and it is able to minimize the uncertainty in the pollutant source. Finally, we provide two case studies to demonstrate efficacy of the proposed motion planning algorithm.

This paper presents a computer vision based system for interaction between multiple humans and multiple robots. Each human can “select” (obtain the undivided attention of) a robot by simply looking directly at it. This extends previous work whereby a single human can select one or more robots from a population. Each robot optimally assigns human identities to tracked faces in its camera view using a local Hungarian algorithm. The gaze-direction and location of the faces are estimated via vision, and a score for each robot-face pair is assigned. Then the system finds the global optimal allocation of robot-to-human selections using a centralized Hungarian algorithm. A useful feature of this method is that robots can be selected by people they cannot see. This is the first demonstration of optimal many-to-many robot-selection HRI.

We consider the problem of self-localisation for a mobile robot in an environment with a requested level of accuracy. The robot moves in a known environment following typical trajectories, which can be characterised in statistical terms. One of the main drivers of this paper is its application to assistive robots guiding senior or impaired users in shopping centres or in other public spaces. To localise itself the robot uses onboard sensors such as encoders and inertial platforms. The level of noise in these sensors and the lack of absolute measurements determines a steady growth of the uncertainty on its position. To alleviate the problem, we assume the presence of a number of visual markers deployed in the environment. Whenever the robot comes across one of these sensors, the uncertainty on its position is reset. In the paper, we show a methodology to minimise the number of these sensors and to select their position so that the uncertainty is never worse than a given target threshold with an assigned probability.

In this paper, we address the problem of planning optimal zero moment point (ZMP) trajectories for the double support phase in bipedal gaits that alternate between single and double support. This is achieved by allowing pre- and post-actuation during the single support phases. Thus, we solve two coupled problems: exact tracking of a given desired ZMP trajectory in the pre- and post-phases (single support), and determination of the desired ZMP during the transition phase (double support). Both are solved while minimizing the overall control energy. We also provide a formal method to assess how the choice of desired ZMP trajectory during the single support phases impacts the overall energy expended during the footstep cycle. Although the obtained solution may not be physically feasible in general, it represents a benchmark to which alternative feasible solutions may be compared. Our approach generalizes previous results that consider only constant output in the pre- and post-phases e.g., allowing pre- and post-phase output from a family of polynomial splines. We evaluate the approach via simulations.

Inertial reorientation of airborne articulated bodies has been an active area of research in the robotics community, as this behavior can help guide dynamic robots to a safe landing with minimal damage. The main objective of this work is emulating the aggressive and large angle correction maneuvers, like somersaults, that are performed by human divers. To this end, a planar three link robot, called DiverBot, is proposed. By considering a gravity-free scenario, a local connection is obtained between joint angles and the body orientation, resulting in a reduction in the system dynamics. An optimal control policy applied on this reduced configuration space yielded diving maneuvers that are dynamically feasible. Numerical results show that the DiverBot can execute one somersault without drift and multiple somersaults with minimal drift.

Optimal control is a powerful approach to achieve optimal behavior. However, it typically requires a manual specification of a cost function which often contains several objectives, such as reaching goal positions at different time steps or energy efficiency. Manually trading-off these objectives is often difficult and requires a high engineering effort. In this paper, we present a new approach to specify optimal behavior. We directly specify the desired behavior by a distribution over future states or features of the states. For example, the experimenter could choose to reach certain mean positions with given accuracy/variance at specified time steps. Our approach also unifies optimal control and inverse optimal control in one framework. Given a desired state distribution, we estimate a cost function such that the optimal controller matches the desired distribution. If the desired distribution is estimated from expert demonstrations, our approach performs inverse optimal control. We evaluate our approach on several optimal and inverse optimal control tasks on non-linear systems using incremental linearizations similar to differential dynamic programming approaches.

This work introduces a hybrid visual servoing technique for differentially flat, underactuated systems that is well suited for aggressive dynamics. Standard Position-Based Visual Servoing (PBVS) and Image-Based Visual Servoing (IBVS) approaches for underactuated systems, such as quadrotors, oftentimes do not explicitly ensure that the relevant image features stay in the camera's field of view, especially while the system is performing agile maneuvers. We present a control technique that is designed to mitigate this issue and that results in increased robustness. Given a goal image, we first solve a constrained Perspective-n-Point (PnP) problem to find an equilibrium pose which aligns the camera with the goal. We then formulate the task of navigating to the goal pose as an optimal control problem, where a cost over the resulting image feature tracks along the trajectory is minimized which implicitly keeps features in the field of view over the course of the trajectory. The optimization is performed over a polynomial parametrization of the flat outputs of the system to decrease the dimensionality of the optimization. Simulations and physical experiments are performed with a quadrotor system to benchmark the algorithm's performance against a typical PBVS approach.

For hydraulic robots, modeling of the control flows is a key issue. The conventional modeling methods are based on the Bernoulli equation (i.e., the control flow is a function of the form equation which needs assumption of steady flow. In robotics, since the actual control flow is not always a steady flow, this paper provides a new practical modeling method without the Bernoulli equation. In a word, the proposed modeling method is a systematic version of the industrial modeling in which table representations are accepted and the assumption of steady flow is not needed. First, based on the input-state equation of hydraulic robots, we propose a new matrix representation of control flows via off-line input-output calculations. Second, based on the projection theorem in the matrix space, the optimal table representation is uniquely determined to be implementable. Finally, in comparison with the conventional modeling method with the Bernoulli equation, the effectiveness of the proposed modeling method is confirmed experimentally. The proposed modeling method is applicable to electric robots with complex friction and so on.

This paper presents OpenSwarm, a lightweight easy-to-use open-source operating system. To our knowledge, it is the first operating system designed for and deployed on miniature robots. OpenSwarm operates directly on a robot's microcontroller. It has a memory footprint of 1 kB RAM and 12 kB ROM. OpenSwarm enables a robot to execute multiple processes simultaneously. It provides a hybrid kernel that natively supports preemptive and cooperative scheduling, making it suitable for both computationally intensive and swiftly responsive robotics tasks. OpenSwarm provides hardware abstractions to rapidly develop and test platform-independent code. We show how OpenSwarm can be used to solve a canonical problem in swarm robotics—clustering a collection of dispersed objects. We report experiments, conducted with five e-puck mobile robots, that show that an OpenSwarm implementation performs as good as a hardware-near implementation. The primary goal of OpenSwarm is to make robots with severely constrained hardware more accessible, which may help such systems to be deployed in real-world applications.

We present an online trajectory optimization approach that optimizes a trajectory such that object recognition performance is improved. Inspired by prior work, we formulate the optimization as a derivative-free stochastic optimization, allowing us to express the cost function in an arbitrary way. The cost function is defined such that information acquisition of target objects is improved, while simultaneously moving towards the goal point. We show the evaluation of our approach on a quadrotor platform in simulation as well as on a real robot. The results show that by using an online optimization approach recognition accuracy is greatly improved, but more importantly the optimized trajectory reduces the uncertainty of the posterior class distribution greatly. Hence, verifying that the optimized trajectory collects more valuable information.

This paper presents an online method of simultaneous localization and mapping (SLAM) for estimating the positions of multiple moving sound sources and stationary robots and synchronizing microphone arrays attached to those robots. Since each robot with a microphone array can solely estimate the directions of sound sources, the two-dimensional source positions can be estimated from the source directions estimated by multiple robots using a triangulation method. In addition, sound mixtures can be separated accurately by regarding distributed microphone arrays as one big array. To perform these tasks, some methods have been proposed for localizing and synchronizing microphone arrays. These methods, however, can be used only if a single sound source exists because the time differences of arrival (TDOAs) between microphones are assumed to be directly observed. To overcome this limitation, we propose a unified state-space model that encodes the source and robot positions and the time offsets between microphone arrays in a latent space. Given the TDOAs and directions of arrival (DOAs) estimated by separating observed mixture sounds into source sounds, the latent variables are estimated jointly in an online manner using a FastSLAM2.0 algorithm that can deal with an unknown time-varying number of moving sound sources.

With the increasing number of clinical interventions using needle shaped tools, robotic control of needle insertion procedures has been an active research field for many years. In this work we propose a 3D model of a flexible needle that takes into account tissue deformations in order to predict the needle shape and trajectory when it is inserted using a robotic arm. To account for tissue displacements, we designed a method based on visual feedback that updates the interaction model between the needle and the tissue using an unscented Kalman filter. Results obtained from several needle insertions in a soft tissue phantom showed that the method gives good performance in terms of needle trajectory prediction. This model was also considered in a closed-loop control approach to allow automatic reaching of a target.

This paper addresses time-optimal path-constrained trajectory planning. Given a geometric path for a manipulator, this paper focuses on the selection of the time law along the path. This law minimizes the time required to complete the path and at the same time is consistent with constraints, both at kinematic and dynamic levels. To obtain the optimal law a decision algorithm for the acceleration along the path has been developed. Remarkably, the algorithm is amenable to online implementation, thus allowing for path replanning. An experimental validation on an ABB IRB140 robot is shown.

In this paper we consider an online planning problem for unmanned aerial vehicle (UAV) operations. Specifically, a UAV has the task of reaching a goal from a set of possible goals while minimizing the amount of energy required. Due to unforeseen disturbances, it is possible that initially attractive goals might end up being very expensive during the execution. Thus, two main problems are investigated here: i) how to predict and plan the motion of the UAV at run time to minimize its energy consumption and ii) when to schedule next replanning time to avoid unnecessary periodic re-evaluation executions. Our approach considers a nonlinear model of the system for which a model predictive controller is used to determine the desired control inputs for each possible goal. These control inputs are then used to estimate the energy required to reach the different goals. Finally, a self-triggered scheduling policy determines how long to wait before replanning the goal to aim for. The proposed framework is validated through simulations and experiments in which a quadrotor must choose and reach some goal while being subject to external disturbances.

This paper presents an online self-supervised approach to improve the quality and relevance of input point cloud to a 3D registration algorithm. The suggested method considers the visibility of the model points and learns discriminative appearance of the object under gradual changes. It selectively reduces the amount of information to process by excluding non-visible points of the model and removing outliers from data stream, which results in better alignment between the input data and the model. Thus, by providing a good initial pose, it speeds up the iterative procedure of EM-like optimization for pose estimation (i.e., ICP) to achieve better efficiency and robustness. We compiled a new object dataset of RGBD images under camera motion with ground truth poses of the camera and the objects. We have performed experiments on this dataset and obtained promising results.

In pursuit of increasing the operational tempo of a ground robotics platform in unknown domains, we consider the problem of predicting the distribution of structural state-estimation error due to poorly-modeled platform dynamics as well as environmental effects. Such predictions are a critical component of any modern control approach that utilizes uncertainty information to provide robustness in control design. We use an online learning algorithm based on matrix factorization techniques to fit a statistical model of error that provides enough expressive power to enable prediction directly from motion control signals and low-level visual features. Moreover, we empirically demonstrate that this technique compares favorably to predictors that do not incorporate this information.

One of the biggest challenges in intelligent robotics is to build robots that can learn to use language. To this end, we think that the practical long-term on-line concept/word learning algorithm for robots is a key issue to be addressed. In this paper, we develop an unsupervised on-line learning algorithm that uses Bayesian nonparametrics for categorizing multimodal sensory signals such as audio, visual, and haptic information for robots. The robot uses its physical body to grasp and observe an object from various viewpoints as well as listen to the sound during the observation. The most important property of the proposed framework is to learn multimodal concepts and the language model simultaneously. This mutual learning framework of concepts and language significantly improves both speech recognition and multimodal categorization performances. We conducted a long-term experiment where a human subject interacted with a real robot over 100 hours using 499 objects. Some interesting results of the experiment are discussed in this paper.

The assistance to people during rehabilitation has to be adapted to their needs. Too little help can lead to frustration and stress in the user; an excess of help may lead to low participation and loss of residual skills. Robotic rollators may adapt assistance. The main challenge to cope with this issue is to estimate how much help is needed on the fly, because it depends not only on the person condition, but also on the specific situation that they are negotiating. Clinical scales provide a global condition based estimation, but no local estimator based on punctual needs. Condition also changes in time, so clinical scales need to be recalculated again and again. In this paper we propose a novel approach to estimate users' condition in a continuous way via a robotic rollator. Our work focuses on predicting the value of the well known Tinetti Mobility test from spatiotemporal gait parameters obtained from our platform while users walk. This prediction provides continuous insight on the condition of the user and could be used to modify the amount of help provided. The proposed method has been validated with 19 volunteers at a local hospital that use a rollator for rehabilitation. All volunteers presented some physical or mental disabilities. Our results successfully show a high correlation of spatiotemporal gait parameters with Tinetti Mobility test gait (R2 = 0.7) and Tinetti Mobility test balance (R2 = 0.6).

One of the key challenges in applying reinforcement learning to complex robotic control tasks is the need to gather large amounts of experience in order to find an effective policy for the task at hand. Model-based reinforcement learning can achieve good sample efficiency, but requires the ability to learn a model of the dynamics that is good enough to learn an effective policy. In this work, we develop a model-based reinforcement learning algorithm that combines prior knowledge from previous tasks with online adaptation of the dynamics model. These two ingredients enable highly sample-efficient learning even in regimes where estimating the true dynamics is very difficult, since the online model adaptation allows the method to locally compensate for unmodeled variation in the dynamics. We encode the prior experience into a neural network dynamics model, adapt it online by progressively refitting a local linear model of the dynamics, and use model predictive control to plan under these dynamics. Our experimental results show that this approach can be used to solve a variety of complex robotic manipulation tasks in just a single attempt, using prior data from other manipulation behaviors.