IROS 2015 Papers

This paper focuses on various control strategies for a modular cable-articulated parallel robotic manipulator called PACER (Parallel Articulated-Cable Exercise Robot). The proposed robotic device is comprised of multiple cables and a single antagonistically actuated prismatic joint which connects a six degrees-of-freedom moving platform to a fixed base. This novel design provides an attractive architecture for implementation of a home-based rehabilitation device as an alternative to expensive serial robots. In this paper, after deriving the dynamic equations via Newton-Euler formulation, the development of different control strategies in the context of upper limb rehabilitation in three-dimensional space will be examined. The various control modes include: (i) passive mode through feedback linearization for recovering functionalities of the joints at the first step of post-stroke rehabilitation, (ii) active mode via admittance controller for assisting patient to do tasks better, and finally (iii) resistive mode via stiffness controller for increasing the muscle strength. This is now evaluated via a simulation case-study and development of a physical testbed is underway. Since the human arm consists of seven degrees-of-freedom, and daily activities of upper limbs are combination of motions of these simple joints, this proposed design with various control strategies has enormous potential for neurological remapping and brain plasticity in post-stroke patients with upper limb injuries.

The deformable and continuum nature of soft robots promises versatility and adaptability. However, control of modular, multi-limbed soft robots for terrestrial locomotion is challenging due to the complex robot structure, actuator mechanics and robot-environment interaction. Traditionally, soft robot control is performed by modeling kinematics using exact geometric equations and finite element analysis.

Physical interaction in robotics is a complex problem that requires not only accurate reproduction of the kinematic trajectories but also of the forces and torques exhibited during the movement. We base our approach on Movement Primitives (MP), as MPs provide a framework for modelling complex movements and introduce useful operations on the movements, such as generalization to novel situations, time scaling, and others. Usually, MPs are trained with imitation learning, where an expert demonstrates the trajectories. However, MPs used in physical interaction either require additional learning approaches, e.g., reinforcement learning, or are based on handcrafted solutions. Our goal is to learn and generate movements for physical interaction that are learned with imitation learning, from a small set of demonstrated trajectories. The Probabilistic Movement Primitives (ProMPs) framework is a recent MP approach that introduces beneficial properties, such as combination and blending of MPs, and represents the correlations present in the movement. The ProMPs provides a variable stiffness controller that reproduces the movement but it requires a dynamics model of the system. Learning such a model is not a trivial task, and, therefore, we introduce the model-free ProMPs, that are learning jointly the movement and the necessary actions from a few demonstrations. We derive a variable stiffness controller analytically. We further extent the ProMPs to include force and torque signals, necessary for physical interaction. We evaluate our approach in simulated and real robot tasks.

This paper presents a non-linear, dynamic model of the flexure-based transmission in the Harvard Ambulatory Microrobot (HAMR). The model is derived from first principles and has led to a more comprehensive understanding of the components in this transmission. In particular, an empirical model of the dynamic properties of the compliant Kapton flexures is developed and verified against theoretical results from beam and vibration theory. Furthermore, the fabrication of the piezoelectric bending actuators that drive the transmission is improved to match theoretical performance predictions. The transmission model is validated against experimental data taken on HAMR for the quasi-static (1–10 Hz) operating mode, and is used to redesign the transmission for improved performance in this regime. The model based redesign results in a 266% increase in the work done by the foot when compared to a previous version of HAMR. This leads to a payload capacity of 2.9g, which is ∼ 2× the robot's mass and a 114% increase. Finally, the model is validated in the dynamic regime (40–150 Hz) and the merits of a second order linear approximation are discussed.

In this paper we present a new method to control instantaneously MIMO oscillating systems. The algorithm and its hardware configuration are implemented on a multidimensional resonant force sensor. The robotic applications of this sensor require real-time resonant frequency measures. Inspired by the technique used in single quartz crystal oscillator, the method is extended to the MIMO case and is presented as an alternative solution to control multiple resonances of a force sensor. Three frequencies of one single active plate, on which several piezoelectric patches are bonded, are simultaneously excited. The method uses spatial modal decoupling technique implemented with analogue electronics thanks to a smart arrangement of the piezoelectric patches. This allows the full MIMO system to be transformed into three independent SISO subsystems. Then, for each of them, a controller is tuned to reinforce self-oscillation of the force sensor system to several resonant working points in a closed-loop configuration. Finally, experimental verification shows that the method can be successfully used for the control of a new type of resonant force sensors.

In recent years, the demand for rocket launching has increased due to the development of space technology. However, using inexpensive rockets is not always possible. Although the cost of solid-propellant rockets is relatively reasonable, safely manufacturing a large amount of solid propellant is difficult, and the manufacturing process is disjointed. Therefore, safe and continues manufacturing of solid propellant is necessary. On the basis of the movements of the intestinal tract, we realized that the movements required for transport and mixing of solid propellants are possible to achieve without the application of a large force. By mimicking these intestinal movements, we can safely and continuously manufacture a large amount of solid propellant. We developed a peristaltic pump, based on bowel peristalsis, using straight-fiber-type artificial muscle, and in this study, we demonstrate the usefulness of this pump for transporting the highly viscous fluids and solid-liquid mixed fluid. We consider that the peristaltic pump can also be effectively used as a mixing device. In this paper, we show that mixing one highly viscous fluid with another is possible by measuring the luminance values. In the manufacturing process, we use glass beads with the same diameter as microparticles and a sodium polyacrylate aqueous solution resembling a high viscosity fluid or hydroxyl-terminated polybutadiene as the raw material in the solid propellant. We then measure the glass beads in the resulting solid-liquid mixed fluid.

Mixed Reality can be a valuable tool for research and development in robotics. In this work, we refine the definition of Mixed Reality to accommodate seamless interaction between physical and virtual objects in any number of physical or virtual environments. In particular, we show that Mixed Reality can reduce the gap between simulation and implementation by enabling the prototyping of algorithms on a combination of physical and virtual objects, including robots, sensors, and humans. Robots can be enhanced with additional virtual capabilities, or can interact with humans without sharing physical space. We demonstrate Mixed Reality with three representative experiments, each of which highlights the advantages of our approach. We also provide a testbed for Mixed Reality with three different virtual robotics environments in combination with the Crazyflie 2.0 quadcopter.

A precise calibration method for optical high-speed gaze controllers using rotational mirrors is proposed. The mirror-based pan/tilt camera can change gaze direction more quickly than a conventional pan-tilt-zoom camera, which actuates the camera itself. The high-speed gaze controller can track moving objects at high magnification and can utilize new applications of three-dimensional measurement. However, the controller is difficult to calibrate owing to shallow depth of field (DOF) at close focus positions and mechanical differences. We propose to control optics and illumination to detect calibration pattern control points precisely against shallow DOF; we deepen the DOF using a small aperture, coaxial episcopic illumination, and a retroreflective checkerboard pattern. We introduce mirror thickness for an exact gaze control model and bundle adjustment for large center-of-projection translation. We confirm that wide gaze range calibration is achieved more accurately than when mirror thickness is not considered.

Flexible serpentine manipulators are widely used in surgical robots as it can be operated inside the patient's body cavity by backbone bending. However, during the bending the manipulator sweeps over a region, where sensitive organs may locate. This raises the safety concern. In this paper, a motion planning algorithm focusing on minimize the sweeping area for flexible serpentine manipulators is presented. Particularly, a three dimensional backward average neural dynamic model (BANDM) is proposed to build minimum sweeping area planning field in the configuration space of the serpentine manipulator. Given a target position, the motion sequence is generated automatically based on the established planning field. The simulations and experimental results validate the effectiveness and superiority of the proposed planning approach over conventional planning algorithms in terms of sweeping area with keeping target reach and obstacle avoidance.

We consider the problem of computing patrolling strategies under communication constraints for a team of autonomous robots employed in repeated surveillance missions on a set of predefined locations. We assume the presence of a communication infrastructure providing only some regions of the environment with a communication link to a mission control center (MCC). We define the problem of computing a joint patrolling strategy that minimizes communication latencies, defined as the delays between inspecting some locations and reporting the outcome to the MCC. We provide and experimentally evaluate a MILP formulation and a heuristic method.

This paper presents a new soft-bodied millimeterscale swimmer actuated by rotating uniform magnetic fields. The proposed swimmer moves through internal undulatory deformations, resulting from a magnetization profile programmed into its body. To understand the motion of the swimmer, a mathematical model is developed to describe the general relationship between the deflection of a flexible strip and its magnetization profile. As a special case, the situation of the swimmer on the water surface is analyzed and predictions made by the model are experimentally verified. Experimental results show the controllability of the proposed swimmer under a computer vision-based closed-loop controller. The swimmers have nominal dimensions of 1.5×4.9×0.06 mm and a top speed of 50 mm/s (10 body lengths per second). Waypoint following and multiagent control are demonstrated for swimmers constrained at the air-water interface and underwater swimming is also shown, suggesting the promising potential of this type of swimmer in biomedical and microfluidic applications.

This paper presents a new control method devoted to improve the performances of 2-DOF piezoelectric actuator for precise positioning tasks. The piezoelectric actuator exhibits strong cross-couplings between its two axes and strong hysteresis and creep nonlinearities. These unwanted phenomena undeniably compromised therefore the final precision of the targeted tasks (micromanipulation) and should be conveniently accounted during the controller synthesis. In this paper, we proposed a combination of the discrete linear Kalman filtering with a closed-loop scheme to suppress the effects of the couplings and of the nonlinearities. The suggested method permits to improve the performances of the piezoelectric actuator without specific and detailed characterization and knowledges on the hysteresis and on the creep models. Extensive experiments were carried out with complex desired trajectories and demonstrate the efficiency of the novel approach.

3D posture estimation for a hose-shaped robot is critical in rescue activities due to complex physical environments. Conventional sound-based posture estimation assumes rather flat physical environments and focuses only on 2D, resulting in poor performance in real world environments with rubble. This paper presents novel 3D posture estimation by exploiting microphones and accelerometers. The idea of our method is to compensate the lack of posture information obtained by sound-based time-difference-of arrival (TDOA) with the tilt information obtained from accelerometers. This compensation is formulated as a nonlinear state-space model and solved by the unscented Kalman filter. Experiments are conducted by using a 3m hose-shaped robot with eight units of a microphone and an accelerometer and seven units of a loudspeaker and a vibration motor deployed in a simple 3D structure. Experimental results demonstrate that our method reduces the errors of initial states to about 20 cm in the 3D space. If the initial errors of initial states are less than 20 %, our method can estimate the correct 3D posture in real-time.

Performance benchmarking has become an important topic within robotics. It is indeed, a critical way to compare different solutions under different conditions. In this paper, we focus on performance benchmarking of multi-robot systems which explore and map unknown terrains. We present a collection of metrics to objectively compare different algorithms that can be applied to collaborative multi-robot exploration. We also identify parameters that impact robotic fleet performances. By varying the parameters, we can identify strengths and limits of an algorithm. This work is also a first concrete step to address the general problem of objectively comparing different multi-robot coordination algorithms. We illustrate these contributions with realistic simulations of the frontier-based exploration strategy. The simulations were implemented in ROS, which enables to uncouple the control software from the drivers of the robot body. We can therefore use the same code on both simulation and real robots.

Accurate metrical localization is one of the central challenges in mobile robotics. Many existing methods aim at localizing after building a map with the robot. In this paper, we present a novel approach that instead uses geo-tagged panoramas from the Google Street View as a source of global positioning. We model the problem of localization as a non-linear least squares estimation in two phases. The first estimates the 3D position of tracked feature points from short monocular camera sequences. The second computes the rigid body transformation between the Street View panoramas and the estimated points. The only input of this approach is a stream of monocular camera images and odometry estimates. We quantified the accuracy of the method by running the approach on a robotic platform in a parking lot by using visual fiducials as ground truth. Additionally, we applied the approach in the context of personal localization in a real urban scenario by using data from a Google Tango tablet.

This paper presents a definition of convexity useful for describing local optimality in configuration spaces, proves that finding convex regions is relatively easy, and presents an algorithm for approximating the free configuration space using a set of such convex regions. The paper examines simple but interesting systems: serial planar arms with revolute joints, and a Reeds-Shepp car. The paper experimentally explores an approach for finding good (although not necessarily optimal) trajectories using the derived data structure.

Robot systems used in surgical procedures can autonomously position tools at points correlated with preoperative imaging techniques such as magnetic resonance (MR) and computed tomography (CT). The aim of this paper is to measure and assess medical applicability of a low-cost, lightweight industrial robot arm (Universal robot UR5) guided with the medically certified optical tracking system (Polaris Vicra) to positions registered from a CT scan. Technical setup, measurement equipment, device communication and robot control based on OTS feedback are described. Robot intrinsic accuracy, CT scan accuracy and two methods of robot tool positioning with aid of the optical tracking system (OTS) are measured. Measurements show RMS error of the robot (0.669 mm) is decreased 55.4% when guided with OTS using a single marker probe (0.29 mm) and 40.5% when using OTS with relative referencing (0.39 mm). RMS error of the CT scan readings is 0.46 mm.

Cyanobacteria have ion channels which response to osmolarity change in living environment. Since it may lead to the mechanical characteristics changes of the cell itself due to the intracellular pressure changes, we focus on constructing the mechanical characteristics measurement system of Synechocystis sp. PCC 6803, one of unicellular cyanobacteria. There are two important issues in constructing the system. Because of features of the cyanobacteria, which is a kind of floating cell whose diameter is a few micrometer, one is how to transport the cell to the measurement point. The other is how to deform the target cell. In this paper, we propose the system combining optical tweezers for transportation of the target cell with the robot integrated microfluidic chip (robochip). The robochip contains an on-chip probe actuated by an external actuator, and a force sensor. We present the concept of the total system, and demonstrated the results of mechanical characterization of a single cyanobacteria.

Robust fingertip force detection from fingernail image is a critical strategy that can be applied in many areas. However, prior research fixed many variables that influence the finger color change. This paper analyzes the effect of the finger joint on the force detection in order to deal with the constrained finger position setting. A force estimator method is designed: a model to predict the fingertip force from finger joints measured from 2D cameras and 3 rectangular markers in cooperation with the fingernail images are trained. Then the error caused by the color changes of the joint bending can be avoided. This strategy is a significant step forward from a finger force estimator that requires tedious finger joint setting. The approach is evaluated experimentally. The result shows that it increases the accuracy over 10% for the force in conditions of the finger joint free movement. The estimator is used to demonstrate lifting and replacing objects with various weights.