ICRA 2018 Papers

As legged robots are sent into unstructured environments, the ability to robustly manage contact transitions will be a critical skill. This paper introduces an approach to probabilistically fuse contact models, managing uncertainty in terrain geometry, dynamic modeling, and kinematics to improve the robustness of contact initiation at touchdown. A discrete-time extension of the generalized-momentum disturbance observer is presented to increase the accuracy of proprioceptive force control estimates. This information is fused with other contact priors under a framework of Kalman Filtering to increase robustness of the method. This approach results in accurate contact detection with 99.3 % accuracy and a small 4-5ms delay. Using this new detector, an Event-Based Finite State Machine is implemented to deal with unexpected early and late contacts. This allows the robot to traverse cluttered environments by modifying the control actions for each individual leg based on the estimated contact state rather than adhering to a rigid time schedule regardless of actual contact state. Experiments with the MIT Cheetah 3 robot show the success of both the detection algorithm, as well as the Event-Based FSM while making unexpected contacts during trotting.

Fabrication of vascular network is an important research for transporting required nutrients and oxygen to the artificial tissues. In this paper, we propose a novel method to construct a hepatic lobule-like vascular network in a 3D cellular structure. The network is simply constructed by three types of veins, central vein, portal vein, and sinusoids. To realize these kinds veins, we utilize two different sizes of steel rods and magnetic fibers for delivering nutrients in 3D cellular structure. Alginate gel fibers embedding ferrite particles are prepared as the same length and are magnetized by magnetizer at 3T. A magnetic tweezer with seven poles is proposed to generate sufficient forces that can manipulate magnetized fibers. Here, two types of rods are magnetized to different magnetic poles in order to attract opposite the end of fibers. This manipulation process is performed in fibrinogen and thrombin solution with liver cells (RLC-18). After solidification of the solution, we deposit solutions with cells and fibers repeatedly, and therefore, a multi-layered structure can be constructed. In addition, we investigate cell a viability in fibrin gel according to the depth of the gel. The result is that the deeper the depth of the gel is, the lower the cell viability is. The cell viability is conducted in several condition. As a result, at the low temperature (here at 22 °C), the viability of cell is increased.

We present a new paradigm for real-time object-oriented SLAM with a monocular camera. Contrary to previous approaches, that rely on object-level models, we construct category-level models from CAD collections which are now widely available. To alleviate the need for huge amounts of labeled data, we develop a rendering pipeline that enables synthesis of large datasets from a limited amount of manually labeled data. Using data thus synthesized, we learn category-level models for object deformations in 3D, as well as discriminative object features in 2D. These category models are instance-independent and aid in the design of object landmark observations that can be incorporated into a generic monocular SLAM framework. Where typical object-SLAM approaches usually solve only for object and camera poses, we also estimate object shape on-the-fty, allowing for a wide range of objects from the category to be present in the scene. Moreover, since our 2D object features are learned discriminatively, the proposed object-SLAM system succeeds in several scenarios where sparse feature-based monocular SLAM fails due to insufficient features or parallax. Also, the proposed category-models help in object instance retrieval, useful for Augmented Reality (AR) applications. We evaluate the proposed framework on multiple challenging real-world scenes and show - to the best of our knowledge - first results of an instance-independent monocular object-SLAM system and the benefits it enjoys over feature-based SLAM methods.

This paper addresses a fundamental question of multi-agent knowledge distribution: what information should be sent to whom and when, with the limited resources available to each agent? Intelligent Knowledge Distribution is a framework that answers these questions. Communication requirements for multi-agent systems can be rather high when an accurate picture of the environment and the state of other agents must be maintained. To reduce the impact of multi-agent coordination on systems, including communications, this paper introduces the concept of action-based constraints on partially observable Markov decision processes, rewards based upon the value of information driven by Kullback-Leibler Divergence, and probabilistic constraint satisfaction through discrete optimization and Markov chain Monte Carlo analysis. Intelligent Knowledge Distribution is driven by determining the information content an agent believes another agent will obtain by receiving certain information, along with the importance or relevance of that information to the system objective. To perform constraint analysis on an infinite-horizon policy, policies are represented as a Finite State Controller allowing Markov chain Monte Carlo analysis to determine a probabilistic level of guarantee that the constraints will be satisfied. The analysis of performance for an example mission presented in this paper shows the constrained controllers, during the highest constraint seen in simulations, can be constructed to meet minimal constraint guarantees (80%) while impacting the optimal value less than 50%, where the unconstrained optimal controller only satisfied the constraint 10% of the time.

This paper presents a novel constrained, sampling-based motion planning method for grasp and transport tasks with a redundant robotic manipulator. We utilize a planning margin for grasping with constraints that allow the best grasp configuration and approach direction to be determined automatically. For manipulators with many degrees of freedom, our method efficiently chooses the optimal grasp pose when there are many redundant solutions. The method also introduces a parameterized intermediate pose that is optimized to determine the approach direction, increasing robustness under sensor uncertainty and execution errors. Our method also considers transporting the grasped object to the desired target position using a Rapidly-exploring Random Tree (RRT) algorithm that incorporates soft constraints via appropriate cost penalties. We demonstrate the effectiveness and efficiency of our algorithms on a number of simulated and experimental applications. Our experimental results show a marked improvement in computational efficiency in comparison to previously studied approaches.

Tracking planar objects has a wide range of applications in robotics. Conventional template tracking algorithms, however, often fail to observe fast object motion or drift significantly after a period of time, due to drastic object appearance change. To address such challenges, we propose a novel constrained confidence matching algorithm for motion estimation and a robust Kalman filter for template updating. Integrated with an accurate occlusion detector, our approach achieves accurate motion estimation in presence of partial occlusion, by excluding occluded pixels from computation of motion parameters. Furthermore, the proposed Kalman filter employs a novel control-input model to handle the object appearance change, which brings our tracker high robustness against sudden illumination change and heavy motion blur. For evaluation, we compare the proposed tracker with several state-of-the-art planar object trackers on two public benchmark datasets. Experimental results show that our algorithm achieves robust tracking results against various environmental variations, and outperforms baseline algorithms remarkably on both datasets.

Sensor guided, automated systems require the composition of various sensors and data processing algorithms to obtain relevant information for performing their task. Many applications have additional requirements such as a certain accuracy, which has to be achieved despite sensor noise and calibration errors. In this paper we model the configuration of perception systems as a planning problem over probabilistic graphical models. We work on a subset of the full configuration space of perceptions systems, specifically the used sensors, data processing algorithms and view poses. Based on a semantic description of the goal, available sensors and data processing algorithms, our system plans perception steps and sensor data fusion autonomously. The planner operates by constructing a factor graph until the accuracy requirements of tasks are fulfilled or unobtainable with the available action set. We validate our approach in an industrial assembly scenario.

In this paper, we consider the problem of data association in feature cloud matching. While Joint Compatibility (JC) test is a widely adopted technique for searching the global optimal data association, it becomes less restrictive as more features are well matched. The early well-matched features contribute little to total matching cost while the gating threshold increases in the chi-square test, which allows the acceptance of bad feature pairings in the last step. In this paper, we propose the Conditional Compatibility (CC) test, which is not only more restrictive than JC test, but also probabilistically sound. The proposed test of a new feature pairing is based on the conditional probability distribution of feature locations given the early pairings. CC test can be added into any JC test based search algorithm, such as Joint Compatibility Branch and Bound (JCBB), Incremental Posterior Joint Compatibility (IPJC) and FastJCBB, without increasing much computational complexity. The more restrictive criterion of accepting a feature pairing, not only helps to reject bad associations, but also bounds the search space, which substantially improves the search efficiency. The real matching experiments justify that our algorithm produces better feature cloud matching results in a more efficient manner.

Model-free deep reinforcement learning has been shown to exhibit good performance in domains ranging from video games to simulated robotic manipulation and locomotion. However, model-free methods are known to perform poorly when the interaction time with the environment is limited, as is the case for most real-world robotic tasks. In this paper, we study how maximum entropy policies trained using soft Q-learning can be applied to real-world robotic manipulation. The application of this method to real-world manipulation is facilitated by two important features of soft Q-learning. First, soft Q-learning can learn multimodal exploration strategies by learning policies represented by expressive energy-based models. Second, we show that policies learned with soft Q-learning can be composed to create new policies, and that the optimality of the resulting policy can be bounded in terms of the divergence between the composed policies. This compositionality provides an especially valuable tool for real-world manipulation, where constructing new policies by composing existing skills can provide a large gain in efficiency over training from scratch. Our experimental evaluation demonstrates that soft Q-learning is substantially more sample efficient than prior model-free deep reinforcement learning methods, and that compositionality can be performed for both simulated and real-world tasks.

Compliant motions allow alignment of workpieces using naturally occurring interaction forces. However, free-floating objects do not have a fixed base to absorb the reaction forces caused by the interactions. Consequently, if the interaction forces are too high, objects can gain momentum and move away after contact. This paper proposes an approach based on direct force control for compliant manipulation of free-floating objects. The objective of the controller is to minimize the interaction forces while maintaining the contact. The proposed approach achieves this by maintaining small constant force along the motion direction and an apparent reduction of manipulator inertia along remaining Degrees of Freedom (DOF). Simulation results emphasize the importance of relative inertia of the robotic manipulator with respect to the free-floating object. The experiments were performed with KUKA LWR4+ manipulator arm and a two-dimensional micro-gravity emulator (object floating on an air bed), which was developed in-house. It was verified that the proposed control law is capable of controlling the interaction forces and aligning the tools without pushing the object away. We conclude that direct force control works better with a free-floating object than implicit force control algorithms, such as impedance control.

The development of soft, compliant force sensors is greatly sought after in areas such as soft robotics and prosthetics. Nevertheless, solutions for measuring forces in multiple axes, while being mechanically compliant, have been few and far between. Here we present a compliant sensor able to detect forces tangential and normal to the sensor surface. The transduction mechanism is based on the deformation of laser-machined carbon fiber composite (CFC) micro-scale meanders, encapsulated within elastomers layers. Strains in the elastomer are transmitted to the meanders, causing changes in the electrical resistance of the sensor contact mechanics. Configuring the meanders in a radial pattern, segmenting them into quadrants (two antagonist pairs) and biasing the center of the sensor out-of-plane enables detection of forces in multiple axis via differential measurement. Sensors were manufactured using a custom fabrication process and exhibited high mechanical compliance with a very low form factor. The sensors were experimentally characterized and demonstrated large differential changes in resistance (up to 26 kΩ for tangential forces applied to the sensor surface). We integrated our sensor onto a soft robotic gripper finger and demonstrated the ability to detect changes in friction at the actuator surface, thus demonstrating their potential for real world applications.

Sparsity has been widely recognized as crucial for efficient optimization in graph-based SLAM. Because the sparsity and structure of the SLAM graph reflect the set of incorporated measurements, many methods for sparsification have been proposed in hopes of reducing computation. These methods often focus narrowly on reducing edge count without regard for structure at a global level. Such structurally-naïve techniques can fail to produce significant computational savings, even after aggressive pruning. In contrast, simple heuristics such as measurement decimation and keyframing are known empirically to produce significant computation reductions. To demonstrate why, we propose a quantitative metric called elimination complexity (EC) that bridges the existing analytic gap between graph structure and computation. EC quantifies the complexity of the primary computational bottleneck: the factorization step of a Gauss-Newton iteration. Using this metric, we show rigorously that decimation and keyframing impose favorable global structures and therefore achieve computation reductions on the order of r2/9 and r3, respectively, where r is the pruning rate. We additionally present numerical results showing EC provides a good approximation of computation in both batch and incremental (iSAM2) optimization and demonstrate that pruning methods promoting globally-efficient structure outperform those that do not.

This paper presents a Light Detection and Ranging (LiDAR) data set that targets complex urban environments. Urban environments with high-rise buildings and congested traffic pose a significant challenge for many robotics applications. The presented data set is unique in the sense it is able to capture the genuine features of an urban environment (e.g. metropolitan areas, large building complexes and underground parking lots). Data of two-dimensional (2D) and three-dimensional (3D) LiDAR, which are typical types of LiDAR sensors, are provided in the data set. The two 16-ray 3D LiDARs are tilted on both sides for maximal coverage. One 2D LiDAR faces backward while the other faces forwards to collect data of roads and buildings, respectively. Raw sensor data from Fiber Optic Gyro (FOG), Inertial Measurement Unit (IMU), and the Global Positioning System (GPS) are presented in a file format for vehicle pose estimation. The pose information of the vehicle estimated at 100 Hz is also presented after applying the graph simultaneous localization and mapping (SLAM) algorithm. For the convenience of development, the file player and data viewer in Robot Operating System (ROS) environment were also released via the web page. The full data sets are available at: http://irap.kaist.ac.kr/dataset. In this website, 3D preview of each data set is provided using WebGL.

This paper analyzes the completion time of automated manufacturing systems, especially a dual-armed cluster tool, equipped with parallel processing modules (PMs). Cluster tools, which consist of multiple PMs, a transport robot, and loadlocks where wafer lots are loaded and unloaded, perform semiconductor manufacturing processes, such as lithography, etching, deposition, and testing. Wafer lots are transported by overhead hoist transports (OHTs) between tools or stockers where wafer lots are stored. To reduce the idle time of OHTs or cluster tools, it is essential to estimate the time when all wafers of a lot finish processing in a tool. Hence, we derive closed-form expressions for the completion time of wafer lots, especially in dual-armed cluster tools with parallel PMs to reflect real circumstances of fabs. We finally show that the formulas derived can be used even when there are small processing time variations with numerical experiments.

This paper studies bipedal locomotion as a nonlinear optimization problem based on continuous and discrete dynamics, by simultaneously optimizing the remaining step duration, the next step duration and the foot location to achieve robustness. The linear inverted pendulum as the motion model captures the center of mass dynamics and its low-dimensionality makes the problem more tractable. We first formulate a holistic approach to search for optimality in the three-dimensional parametric space and use these results as baseline. To further improve computational efficiency, our study investigates a sequential approach with two stages of customized optimization that first optimizes the current step duration, and subsequently the duration and location of the next step. The effectiveness of both approaches is successfully demonstrated in simulation by applying different perturbations. The comparison study shows that these two approaches find mostly the same optimal solutions, but the latter requires considerably less computational time, which suggests that the proposed sequential approach is well suited for real-time implementation with a minor trade-off in optimality.

Duchenne muscular dystrophy leaves patients with severe dependency on health care. In an effort to increase independence and quality of life, active exoskeletons are developed to support activities of daily living. This study is dedicated to the development and assessment of three different admittance control algorithms for a trunk supporting robot; a law with constant parameters, a law with added feed-forward force, and a law with variable parameters. A Fitts'-like experiment with 12 healthy subjects was performed to compare the control laws. The results show decreased movement times for the feedforward and variable admittance controllers with respect to the standard admittance.

Applying robots in narrow and cluttered disaster environments such as oil refineries requires a slim body and a wide range of motion. It is also necessary to have abilities to absorb unexpected contact with the environment and to walk on scattered debris. In this paper we propose new compact and high performance torque-controlled actuators for legged robots to satisfy the above mentioned requirements. For axial compactness, torque sensors are designed as ring-shaped thin cylinders surrounding motors or gears with strain gauges for sensing. To achieve broad bandwidth of torque control, we introduced an analog differentiator circuit into an analog digital converter (ADC) board in order to suppress noise in the differential control of joint torque. We also propose methods to reduce torque ripple caused by the deformation of the harmonic drive gear and electromagnetic interference (EMI) from a motor and a motor driver. Finally, experiments of a collision with objects and movement on scattered debris were executed with a fully torque-controlled legged robot built with the proposed actuators.

This paper introduces a novel approach for sensing the bending deformation on soft robots by leveraging multicolor 3D printing. The measurement of deformation enables to complete the feedback loop of deformation control on soft actuators. The working principle of our approach is based on using compact color sensors to detect deformation that is visualized by the change of color ratios. Two novel designs are presented to generate color signals on 3D printed objects, which we call an external signal generator and an internal signal generator. Signal processing and calibration methods are developed to transform the raw RGB-data into a meaningful deformation metric. Our experimental tests taken on soft pneumatic actuators verify that color signals can be stably generated and captured to indicate the bending deformation. The results also demonstrate the usability of this sensing approach in deformation control.

This paper presents a real-time motion planning and control design of a robotic arm for human-robot collaborative safety. A novel collision-free motion planning method is proposed not only to keep robot body from colliding with objects but also preserve the execution of robot's original task under the Cartesian constraint of the environment. Multiple KinectV2 depth cameras are utilized to model and track dynamic obstacles (e.g. Humans and objects) inside the robot workspace. Depth images are applied to generate point cloud of segmented objects in the environment. A K-nearest neighbor (KNN) searching algorithm is used to cluster and find the closest point from the obstacle to the robot. Then a Kalman filter is applied to estimate the obstacle position and velocity. For the collision avoidance in collaborative operation, attractive and repulsive potential is generated for robot end effector based on the task specification and obstacle observation. Practical experiments show that the 6-DOF robot arm can effectively avoid an obstacle in a constrained environment and complete the original task.