IROS 2016 论文列表

Recently, high resolution visual methods based on direct-phase measurement of periodic patterns has been proposed with successful applications to microrobotics. This paper proposes a new implementation of direct-phase measurement methods to achieve 3-DoF (degrees of freedom) visual servoing. The proposed algorithm relies on a single frequency tracking rather than a complete 2D discrete Fourier transform that was required in previous works. The method does not require any calibration step and has many advantages such as high subpixelic resolution, high robustness and short computation time. Several experimental validations (in favorable and unfavorable conditions of use) were performed using a XYδ microrobotic platform. The obtained results demonstrate the efficiency of the frequency-based controller, this in term of accuracy (micrometric error), convergence rate (30 iterations in nominal conditions) and robustness.

In this paper, we present a unified approach for Expectation Maximization (EM) based motion segmentation, estimation and analysis from dense point cloud data. When identifying an underlying motion, literature mainly focuses on three related topics: motion segmentation, estimation and analysis. These topics are, however, mostly considered separately while integrated approaches are rare. Our approach specifically focuses on analyzing articulated motion from dense point cloud data by simultaneously solving for three topics using an integrated approach. No prior knowledge, such as background regions, number of segments and correspondence, is required since two iterations in this algorithm allow us to seamlessly accomplish integration of the three tasks. The first iteration of the algorithm is performed between segmentation and estimation, followed by the second iteration between motion estimation and analysis. For the first iteration, we propose EM based subspace clustering algorithm. For the second iteration, we simply fuse the motion analysis method from [1] into an iterative motion estimation algorithm. As a result, we can extract label, correspondence and motion of moving objects simultaneously from dense point cloud sequence. In experiment, we validate the performance of the proposed method on both synthetic and real world data.

Recent research in appearance-based mapping has introduced a diverse range of techniques to deal with environments under varying conditions. Common to almost all the existing methods is that they expect a supervised or offline training. Furthermore, some approaches match sequences, assuming constant velocity over the routes. This work addresses the challenges of appearance-based mapping in an online setup without making assumptions or acquiring prior knowledge of the environment. For this purpose, we exploit the topology preserving capability of self-organizing neural networks and learn the perceptual representation of environments using GIST features. Due to the fact that real-world environments are complex and large-scale, we let the network grow while accounting for the amount of error in the network due to perceptual differences among the places. Given the current state of the network and a query image at any time instant, it is possible to identify whether a place comes from the visited location by computing maximum a posteriori estimate over the network. The extensive experiments on three standard datasets, St. Lucia (4 videos), KITTI (2 sequences), and the Oxford City Center dataset, demonstrate the strength of our approach for real-time place recognition while concurrently learning the spatial representation of scenes.

Model-based trajectory optimization often fails to find a reference trajectory for under-actuated bipedal robots performing highly-dynamic, contact-rich tasks in the real world due to inaccurate physical models. In this paper, we propose a complete system that automatically designs a reference trajectory that succeeds on tasks in the real world with a very small number of real world experiments. We adopt existing system identification techniques and show that, with appropriate model parameterization and control optimization, an iterative system identification framework can be effective for designing reference trajectories. We focus on a set of tasks that leverage the momentum transfer strategy to rapidly change the whole-body from an initial configuration to a target configuration by generating large accelerations at the center of mass and switching contacts.

This paper presents a simulation framework for pathological gait assistance with a hip exoskeleton. Previously we had developed an event-driven controller for gait assistance [1]. We now simulate (or optimize) the gait assistance in ankle pathologies (e.g., weak dorsiflexion or plantarflexion). It is done by 1) utilizing the neuromuscular walking model, 2) parameterizing assistive torques for swing and stance legs, and 3) performing dynamic optimizations that takes into account the human-robot interactive dynamics. We evaluate the energy expenditures and walking parameters for the different gait types. Results show that each gait type should have a different assistance strategy comparing with the assistance of normal gait. Although we need further studies about the pathologies, our simulation model is feasible to design the gait assistance for the ankle muscle weaknesses.

This paper deals with the development of a vision-based controller for robot-assisted medical applications. It concerns the use of shearlet coefficients in case of ultrasounds (US) images as visual signal inputs and the design of the associated interaction matrix. The proposed controller was validated in both simulation and on an experimental test bench which consists of a robotic arm holding an US probe in contact with a realistic abdominal phantom. Also, the proposed control scheme was compared to the photometry-based visual servoing approach in order to evaluate its efficiency in different conditions of use (nominal and unfavorable conditions).

We consider the problem of refining an abstract task plan into a motion trajectory. Task and motion planning is a hard problem that is essential to long-horizon mobile manipulation. Many approaches divide the problem into two steps: a search for a task plan and task plan refinement to find a feasible trajectory. We apply sequential quadratic programming to jointly optimize over the parameters in a task plan (e.g., trajectories, grasps, put down locations). We provide two modifications that make our formulation more suitable to task and motion planning. We show how to use movement primitives to reuse previous solutions (and so save optimization effort) without trapping the algorithm in a poor basin of attraction. We also derive an early convergence criterion that lets us quickly detect unsatisfiable constraints so we can re-initialize their variables. We present experiments in a navigation amongst movable objects domain and show substantial improvement in cost over a backtracking refinement algorithm.

This paper presents a technique to efficiently solve the Hamilton-Jacobi-Bellman (HJB) equation for a class of stochastic affine nonlinear dynamical systems in high dimensions. The HJB solution provides a globally optimal controller to the associated dynamical system. However, the curse of dimensionality, commonly found in robotic systems, prevents one from solving the HJB equation naively. This work avoids the curse by representing the linear HJB equation using tensor decomposition. An alternating least squares (ALS) based technique finds an approximate solution to the linear HJB equation. A straightforward implementation of the ALS algorithm results in ill-conditioned matrices that prevent approximation to a high order of accuracy. This work resolves the ill-conditioning issue by computing the solution sequentially and introducing boundary condition rescaling. Both of these additions reduce the condition number of matrices in the ALS-based algorithm. A MATLAB tool, Sequential Alternating Least Squares (SeALS), that implements the new method is developed. The performance of SeALS is illustrated using three engineering examples: an inverted pendulum, a Vertical Takeoff and Landing aircraft, and a quadcopter with state up to twelve.

In this paper, we describe an approach to autonomous navigation using a state observer for sensor based control. In particular, we investigate the task of corridor following in the case of sensory failure (unreliable data) and/or loss of measurement. Our approach uses a virtual sensory frame to project laser rangefinder scans, and builds visual features (namely the position of the vanishing point and the orientation of the corridor median line) to be used in the control law. To insure a safe and smooth navigation even when those parameters cannot be extracted, we design a finite time state observer to estimate the visual features with the objective of maintaining an efficient control of the robot. Simulation and experimental tests validate the proposed approach.

There are many applications where domain-specific sensing, such as accelerometers, kinematics, or force sensing, provide unique and important information for control or for analysis of motion. However, it is not always the case that these sensors can be deployed or accessed beyond laboratory environments. For example, it is possible to instrument humans or robots to measure motion in the laboratory in ways that it is not possible to replicate in the wild. An alternative, which we explore in this paper, is to address situations where accurate sensing is available while training an algorithm, but for which only video is available for deployment. We present two examples of this sensory substitution methodology. The first variation trains a convolutional neural network to regress real-valued signals, including robot end-effector pose, from video. The second example regresses binary signals derived from accelerometer data which signifies when specific objects are in motion. We evaluate these on the JIGSAWS dataset for robotic surgery training assessment and the 50 Salads dataset for modeling complex structured cooking tasks. We evaluate the trained models for video-based action recognition and show that the trained models provide information that is comparable to the sensory signals they replace.

Bellows-like actuators are popular in soft robotic systems. Sensing the movement of these actuators with traditional sensors is challenging. This work proposes and tests a sensing system based on the changing mutual inductance of wire coils on bellows. A method for modeling the changes in mutual inductance between coils of tightly-packed wires is introduced. Changes in mutual inductance are measured through the self-inductance of a circuit made up of the coils in series. The self-inductance of the circuit measures the bellows bend-angle. The experiments show an approximately quadratic relationship between the bend angle and the measured inductance. From a bend angle of 121 ° to −67 ° the inductance of the circuit increases by 19 %. The bias-inducing effects of shear strain, torsional strain, non-uniform bending, and nearby metal are also explored.

We present a semi-direct visual odometry algorithm for a fisheye-stereo camera. In a tracking thread, we simultaneously track oriented patches and estimate the camera pose. In a mapping thread, we estimate the coordinates and surface normal for each new patch to be tracked. Estimation of the surface normals allows us to track patches over a wide variety of viewpoints. In our algorithm, we do not make use of descriptors and robust descriptor matching to find patch correspondences. Instead, we use photoconsistency-based techniques to find patch correspondences. For tracking, we use sparse model-based image alignment to find the relative motion estimate, and feature alignment to find 2D-3D patch correspondences. For mapping, we use plane-sweeping stereo to find matching patches between stereo images. We also implement a state estimator based on the Extended Kalman Filter (EKF) to fuse inertial measurements and relative pose estimates from our visual odometry implementation. We run experiments in two different outdoor environments to validate our algorithm, and discuss the experimental results. Our implementation runs at an average of 42 Hz on a commodity Intel CPU. To the best of our knowledge, there is no other existing semi-direct visual odometry algorithm for a fisheye-stereo camera.

This paper addresses bird song analysis based on semi-automatic annotation. Research in animal behavior, especially with birds, would be aided by automated (or semiautomated) systems that can localize sounds, measure their timing, and identify their source. This is difficult to achieve in real environments where several birds may be singing from different locations and at the same time. Analysis of recordings from the wild has in the past typically required manual annotation. Such annotation is not always accurate or even consistent, as it may vary both within or between observers. Here we propose a system that uses automated methods from robot audition, including sound source detection, localization, separation and identification. In robot audition these technologies have typically been studied separately; combining them often leads to poor performance in real-time application from the wild. We suggest that integration is aided by placing a primary focus on spatial cues, then combining other features within a Bayesian framework. A second problem has been that supervised machine learning methods typically requires a pre-trained model that may require a large training set of annotated labels. We have employed a semi-automatic annotation approach that requires much less pre-annotation. Preliminary experiments with recordings of bird songs from the wild revealed that for identification accuracy our system outperformed a method based on conventional robot audition.

We introduce a framework for learning regrasping behaviors based on tactile data. First, we present a grasp stability predictor that uses spatio-temporal tactile features collected from the early-object-lifting phase to predict the grasp outcome with a high accuracy. Next, the trained predictor is used to supervise and provide feedback to a reinforcement learning algorithm that learns the required grasp adjustments based on tactile feedback. Our results gathered over more than 50 hours of real robot experiments indicate that the robot is able to predict the grasp outcome with 93% accuracy. In addition, the robot is able to improve the grasp success rate from 42% when randomly grasping an object to up to 97% when allowed to regrasp the object in case of a predicted failure.

Obstacle detection by monocular vision is challenging because a single camera does not provide a direct measure for absolute distances to objects. A self-supervised learning approach is proposed that combines a camera and a very small short-range proximity sensor to find the relation between the appearance of objects in camera images and their corresponding distances. The method is efficient enough to run real time on a small camera system that can be carried onboard a lightweight MAV of 19 g. The effectiveness of the method is demonstrated by computer simulations and by experiments with the real platform in flight.

We investigated different dense multirotor UAV traffic simulation scenarios in open 2D and 3D space, under realistic environments with the presence of sensor noise, communication delay, limited communication range, limited sensor update rate and finite inertia. We implemented two fundamental self-organized algorithms: one with constant direction and one with constant velocity preference to reach a desired target. We performed evolutionary optimization on both algorithms in five basic traffic scenarios and tested the optimized algorithms under different vehicle densities. We provide optimal algorithm and parameter selection criteria and compare the maximal flux and collision risk of each solution and situation. We found that i) different scenarios and densities require different algorithmic approaches, i.e., UAVs have to behave differently in sparse and dense environments or when they have common or different targets; ii) a slower-is-faster effect is implicitly present in our models, i.e., the maximal flux is achieved at densities where the average speed is far from maximal; iii) communication delay is the most severe destabilizing environmental condition that has a fundamental effect on performance and needs to be taken into account when designing algorithms to be used in real life.

With the recent success of visual features from deep convolutional neural networks (DCNN) in visual robot self-localization, it has become important and practical to address more general self-localization scenarios. In this paper, we address the scenario of self-localization from images with small overlap. We explicitly introduce a localization difficulty index as a decreasing function of view overlap between query and relevant database images and investigate performance versus difficulty for challenging cross-view self-localization tasks. We then reformulate the self-localization as a scalable bag-of-visual-features (BoVF) scene retrieval and present an efficient solution called PCA-NBNN, aiming to facilitate fast and yet discriminative correspondence between partially overlapping images. The proposed approach adopts recent findings in discriminativity preserving encoding of DCNN features using principal component analysis (PCA) and cross-domain scene matching using naive Bayes nearest neighbor distance metric (NBNN). We experimentally demonstrate that the proposed PCA-NBNN framework frequently achieves comparable results to previous DCNN features and that the BoVF model is significantly more efficient. We further address an important alternative scenario of “self-localization from images with NO overlap” and report the result.

We address the important problem of achieving robust and easy-to-deploy visual state estimation for micro aerial vehicles (MAVs) operating in complex environments. We use a sensor suite consisting of multiple cameras and an IMU to maximize perceptual awareness of the surroundings and provide sufficient redundancy against sensor failures. Our approach starts with an online initialization procedure that simultaneously estimates the transformation between each camera and the IMU, as well as the initial velocity and attitude of the platform, without any prior knowledge about the mechanical configuration of the sensor suite. Based on the initial calibrations, a tightly-coupled, optimization-based, generalized multi-camera-inertial fusion method runs onboard the MAV with online camera-IMU calibration refinement and identification of sensor failures. Our approach dynamically configures the system into monocular, stereo, or other multi-camera visual-inertial settings, with their respective perceptual advantages, based on the availability of visual measurements. We show that even under random camera failures, our method can be used for feedback control of the MAVs. We highlight our approach in challenging indoor-outdoor navigation tasks with large variations in vehicle height and speed, scene depth, and illumination.

In this paper, we propose a new method to assemble microstructures with biological cells towards in vitro 3D cellular structures. The proposed assembly method uses self-assembly process of magnetized toroidal microstructures. Biocompatible toroidal hydrogel microstructures are prepared by electrodeposition method, and ferrite particles are put on the fabricated structures using poly-L-lysine (PLL). The microstructures are magnetized by the magnetic field at 3 T, and assembled by the magnetic self-assembly process. Biological cells were encapsulated in the microstructures and cultured to achieve high density of cells inside structures. The magnetized microstructures were assembled automatically. The magnetic force generated from the ferrite embedded microstructures was estimated and compared to the fluid resistance applied to the microstructures. The proposed method can be applied to achieve 3D in vitro cell structures with vascular networks for tissue engineering applications.