IROS 2018 Papers

Legged robots that make use of compliant actuators have demonstrated greater robustness of locomotion than their rigid counterparts. Stiffness and damping are key parameters that characterize the adaptation to perturbations. In this work, by drawing inspirations from controllable compliance and damping in existing soft and bio-inspired legged robots, we propose an approach to design a nonlinear controller for the balancing of humanoid robots with rigid bodies. Existing literature has proposed simplified dynamical models of biped robots in order to predict the timing and placement of swing foot for walking without falling. We further employ the properties of invariance to perturbations in damped harmonic oscillators and formulate continuous feedback control in combination with predictive foot stepping in order to achieve continuous adaptive recoveries of the nominal walking cycle from unexpected physical disturbances. Our method allows asymptotic convergence of the rigid body dynamics to a subspace with the desired energy level. We demonstrate the robustness of the proposed algorithm base on extensive push recovery experiments on a NAO robot on flat terrains.

Minimally invasive arthroscopic surgery is a very challenging procedure that requires the manipulation of instruments in limited intraarticular space using distorted and sometimes uninformative images. Localizing the arthroscope reliably and at all times w.r.t. surrounding tissue is of fundamental importance to prevent unintended injury to patients. However, even highly-trained surgeons can struggle to localize the arthro-scope using poor image feedback. In this paper, we propose and demonstrate for the first time a visual Simultaneous Localisation and Mapping (SLAM) system, termed ArthroSLAM, capable of robustly and reliably localizing an arthroscope inside a human knee joint. The proposed system fuses the information obtained from the arthroscope, an external camera mounted on an arthroscope holder, and the odometry of a robotic arm manipulating the scope, in an Extended Kalman Filter framework. Also for the first time, we implement five alternative strategies for localization and compare them to our method in a realistic setup with a human cadaver knee joint. ArthroSLAM is shown to outperform the alternative strategies under various challenging conditions, localizing reliably and at all times with a mean Relative Pose Error of up to 1.4mm and 0.7°. Additional experiments conducted with degraded odometry data also validate the robustness of the method. An initial evaluation of the sparse map of a knee section computed by our method exhibits good morphological agreement. All results suggest that ArthroSLAM is a viable component for the robotic orthopedic surgical assistant of the future.

Autonomous soaring capability has the potential to significantly increase time aloft for fixed-wing UAVs. In this paper, we introduce ArduSoar, the first soaring controller integrated into a major autopilot software suite for small UAVs. We describe ArduSoar from the algorithmic standpoint, outline its integration with the ArduPlane autopilot, discuss parameter tuning for it, and conduct a series of flight tests on real sUAVs that show ArduSoar's robustness even in highly nonideal atmospheric conditions.

Unsupervised learning techniques, such as Bayesian topic models, are capable of discovering latent structure directly from raw data. These unsupervised models can endow robots with the ability to learn from their observations without human supervision, and then use the learned models for tasks such as autonomous exploration, adaptive sampling, or surveillance. This paper extends single-robot topic models to the domain of multiple robots. The main difficulty of this extension lies in achieving and maintaining global consensus among the unsupervised models learned locally by each robot. This is especially challenging for multi-robot teams operating in communication-constrained environments, such as marine robots. We present a novel approach for multi-robot distributed learning in which each robot maintains a local topic model to categorize its observations and model parameters are shared to achieve global consensus. We apply a combinatorial optimization procedure that combines local robot topic distributions into a globally consistent model based on topic similarity, which we find mitigates topic drift when compared to a baseline approach that matches topics naïvely, We evaluate our methods experimentally by demonstrating multi-robot underwater terrain characterization using simulated missions on real seabed imagery. Our proposed method achieves similar model quality under bandwidth-constraints to that achieved by models that continuously communicate, despite requiring less than one percent of the data transmission needed for continuous communication.

Robust representation of actions and its sequences for complex robotic tasks would transform robot's understanding to execute robotic tasks efficiently. The challenge is to understand action sequences for highly unstructured environments and to represent and construct action and action sequences. In this manuscript, we present a review of literature dealing with representation of action and action sequences for robot task planning and execution. The methodological review was conducted using Google Scholar and IEEE Xplore, searching the specific keywords. This manuscript gives an overview of current approaches for representing action sequences in robotics. We propose a classification of different methodologies used for action sequences representation and describe the most important aspects of the reviewed publications. This review allows the reader to understand several options that do exist in the research community, to represent and deploy such action representations in real robots.

Estimating accurate and reliable fruit and vegetable counts from images in real-world settings, such as orchards, is a challenging problem that has received significant recent attention. Estimating fruit counts before harvest provides useful information for logistics planning. While considerable progress has been made toward fruit detection, estimating the actual counts remains challenging. In practice, fruits are often clustered together. Therefore, methods that only detect fruits fail to offer general solutions to estimate accurate fruit counts. Furthermore, in horticultural studies, rather than a single yield estimate, finer information such as the distribution of the number of apples per cluster is desirable. In this work, we formulate fruit counting from images as a multi-class classification problem and solve it by training a Convolutional Neural Network. We first evaluate the per-image accuracy of our method and compare it with a state of the art method based on Gaussian Mixture Models over four test datasets. Even though the parameters of the Gaussian Mixture Model based method are specifically tuned for each dataset, our network outperforms it in three out of four datasets with a maximum of 94% accuracy. Next, we use the method to estimate the yield for two datasets for which we have ground truth. Our method achieved 96-97% accuracies. For additional details please see our video here: https://www.youtube.com/watch?v=Le0mb5P-SYc.

We propose an appearance-based along-route localization algorithm that relies on robust place recognition by matching image sequences instead of individual frames. Our approach extends state of the art place recognition framework SeqSLAM in several aspects to realize real-time localization along routes for autonomous navigation. First, our method is online in that we only rely on the recently observed image frames. Second, we provide a homing mechanism based on rotations computed from frame matches. And third, we use a more flexible mechanism to search for matching locations, not restricting the search to straight lines in the cost matrix as in SeqSLAM, but allowing for a wide variety of route traversal conditions such as varying velocities or rotational and translational viewpoint differences. We investigate different image preprocessing steps as well as image similarity metrics wrt. their influence on illumination and viewpoint invariance for a more robust place recognition. On a new challenging dataset, recorded in real world experiments with a planetary rover, in the course of a Moon-analogue mission on Sicily's Mount Etna, we show the feasibility of our direct, sequence-based approach to along-route localization.

Loss of thrust is a critical situation for human pilots of fixed-wing aircraft which force them to select a landing site in the nearby range and perform an emergency landing. The time for the landing site selection is limited by the actual altitude of the aircraft, and it may be fatal if the correct decision is not chosen fast enough. Therefore, we propose a novel RRT* -based planning algorithm for finding the safest emergency landing trajectory towards a given set of possible landing sites. Multiple landing sites are evaluated simultaneously during the flight even before any mechanical issue occurs, and the roadmap of possible landing trajectories is updated permanently. Thus, the proposed algorithm has the any-time property and provides the best emergency landing trajectory almost instantly.

This paper presents a novel and feasible path planning technique for a group of unmanned aerial vehicles (DAVs) conducting surface inspection of infrastructure. The ultimate goal is to minimise the travel distance of DAVs while simultaneously avoid obstacles, and maintain altitude constraints as well as the shape of the UAV formation. A multiple-objective optimisation algorithm, called the Angle-encoded Particle Swarm Optimization (θ- PSO) algorithm, is proposed to accelerate the swarm convergence with angular velocity and position being used for the location of particles. The whole formation is modelled as a virtual rigid body and controlled to maintain a desired geometric shape among the paths created while the centroid of the group follows a pre-determined trajectory. Based on the testbed of 3DR Solo drones equipped with a proprietary Mission Planner, and the Internet-of- Things (loT) for multi-directional transmission and reception of data between the DAV s, extensive experiments have been conducted for triangular formation maintenance along a monorail bridge. The results obtained confirm the feasibility and effectiveness of the proposed approach.

We present the first fully spatial hopping gait of a 12 DoF tailed biped driven by only 4 actuators. The control of this physical machine is built up from parallel compositions of controllers for progressively higher DoF extensions of a simple 2 DoF, 1 actuator template. These template dynamics are still not themselves integrable, but a new hybrid averaging analysis yields a conjectured closed form representation of the approximate hopping limit cycle as a function of its physical and control parameters. The resulting insight into the role of the machines kinematic and dynamical design choices affords a redesign leading to the newly achieved behavior.

Developing analytical models of efficient locomotion in biology is one of the most interesting goals in bio- inspired robotics. This paper presents a mathematical framework in order to model one of the most energy efficient locomotion types in flying animals; i.e., thermal soaring. Unlike the legged locomotion, in flying, modeling the environmental effects on animals' behaviors is very important. In doing so, we develop our model by assuming thermals as bubbles of rising air. According to pieces of real evidence, this kind of modeling is more compatible with the nature of thermal soaring. Moreover, we present a simple hybrid control strategy for obtaining the optimal path in order to maximize benefit from the updraft of air-flow. By using this control strategy, the flying robot can plan a path for traveling between thermals without flapping; i.e., energy efficient flying. So as to investigate the compatibility of presented model and controller with reality, we set their parameters based on the biological evidences. As a result, in simulations, it is observed that the generated flying behavior is comparable with the thermal soaring behavior of real birds. This observation provides a confirmation for generality and applicability of the presented approach.

This paper presents a free-space open-loop dynamic response analysis for an MRI -guided magnetically-actuated steerable intra-vascular catheter system. The catheter tip is embedded with a set of current carrying micro-coils. The catheter is directly actuated via the magnetic torques generated on these coils by the magnetic field of the magnetic resonance imaging (MRI)scanner. The relationship between the input current commands and catheter tip deflection angle presents an inherent nonlinearity in the proposed catheter system. The system nonlinearity is analyzed by utilizing a pendulum model. The pendulum model is used to describe the system nonlinearity and to perform an approximate input-output linearization. Then, a black-box system identification approach is performed for frequency response analysis of the linearized dynamics. The optimal estimated model is reduced by observing the modes and considering the Nyquist frequency of the camera system that is used to track the catheter motion. The reduced model is experimentally validated with 3D open-loop Cartesian free-space trajectories. This study paves the way for effective and accurate free-space closed-loop control of the robotic catheter with real-time feedback from MRI guidance in subsequent research.

Micro-assembly is a challenging issue for automation due to particularities of micro-world physics and limitations on sensors. Consequently, most applications are human-operated often with basic joystick-like interfaces. Beside being nonintuitive, these solutions do not provide their users with a meaningful insight into the microworld. This paper proposes a novel intuitive remote handling interface, using a classical hand-held assembly tool as a paradigm. The master device is a portable instrumented tweezers with one active degree of freedom. Its spatial motion, tracked by optical means, controls the slave kinematics while its pinch commands the slave robot's microgripper and provides haptic feedback. Different coupling strategies using position or speed variables are demonstrated.

Motivated by the desire to have robots physically present in human environments, in recent years we have witnessed an emergence of different approaches for learning active compliance. Some of the most compelling solutions exploit a minimal intervention control principle, correcting deviations from a goal only when necessary, and among those who follow this concept, several probabilistic techniques have stood out from the rest. However, these approaches are prone to requiring several task demonstrations for proper gain estimation and to generating unpredictable robot motions in the face of uncertainty. Here we present a Programming by Demonstration approach for uncertainty-aware impedance regulation, aimed at making the robot compliant - and safe to interact with - when the uncertainty about its predicted actions is high. Moreover, we propose a data-efficient strategy, based on the energy observed during demonstrations, to achieve minimal intervention control, when the uncertainty is low. The approach is validated in an experimental scenario, where a human collaboratively moves an object with a 7-DoF torque-controlled robot.

This paper presents a new actuator designed to produce forces under short stroke displacements. Two variants of the prototype have been manufactured using Multi-Material Additive Manufacturing process, based on a flexible origami-inspired architecture. The structure consists of an airtight chamber constituted by rigid plates combined with flexible hinges and surfaces in order to allow the generation of motion. We propose several insights on integration issues such as limited material resistance and maximum range of motion. Both versions of the prototype are then tested to assess their performances for single strokes and cyclic loading.

This paper addresses the optimization-based planning of collision-free motions for a dual-arm robot with kinematic constraints. Such problems arise, for example, when the robot has to move an object with both arms, whereby the two arms and the gripped object form a closed kinematic chain. Such constrained problems are hard to solve with sampling-based planners, because the probability that a random sample satisfies the closure constraint is practically zero. In contrast, the solution of optimization problems with equality constraints is a well-understood field of research. This paper formulates the motion planning task as optimization problem and proposes a numerical solution using the augmented Lagrangian method for handling constraints. The planner is compared to RRTs, CHOMP and TrajOpt on a set of randomly generated problems for a dual-arm robot with twelve degrees of freedom highlighting the advantages of optimization-based planning.

In this paper, we propose an omnidirectional jumper with expanded locomotion capabilities. The mechanisms for four functions-jumping, steering, self-righting and take-off angle adjustment-are designed using only two motors to maximize the jumping performance. Jumping uses the modified active triggering mechanism with one motor. Steering shares this motor and uses the wheel touching the ground. The take-off angle is adjusted by changing the angle between the body and the foot using another motor. Self-righting is possible by utilizing combinations of the movements that occur in the energy storing and angle adjustment processes. With these four functions, the robot is capable of jumping in all directions and can jump anywhere in between the maximum height and maximum distance. It can also jump multiple times by self-righting. The robot, with a mass of 64.4 g, jumps up to 113 cm in vertical height, and 170 cm in horizontal distance. This robot can be deployed to explore various environments. Moreover, the design method to implement more functions than the number of motors can be applied to design other small-scale robots.

This paper introduces alternative behaviors described by the SLIP model when it is subject to a range of initial conditions. A non-dimensional SLIP model and a numerical return map search scheme are used to determine fixed points as a function of non-dimensional leg stiffness and vertical displacement under friction constraints. A SLIP model behavior analysis is performed, using an analytical stance phase approximation, by diverging from the fixed points, i.e. by increasing/decreasing initial horizontal velocity, and/or touchdown angle. The results show that beyond the regular fixed points, the SLIP model performs an alternative, stable behavior that repeats itself every two cycles of motion. We call these 2nd-order fixed points and the regular ones 1st-order fixed points. A numerical simulation scheme was developed to investigate 2nd-order fixed points for a wide range of horizontal velocities and touchdown angles. Results show that 2nd-order fixed points respecting the friction cone constraints exist that can lead to a number of different behaviors such as high jumps, obstacle avoidance of different heights, or backward motion.

In this paper we study the distance-based docking problem of unmanned aerial vehicles (UAVs) by using a single landmark placed at an arbitrarily unknown position. To solve the problem, we propose an integrated estimation-control scheme to simultaneously achieve the relative localization and navigation tasks for discrete-time integrators under bounded velocity: a nonlinear adaptive estimation scheme to estimate the relative position to the landmark, and a delicate control scheme to ensure both the convergence of the estimation and the asymptotic docking at the given landmark. A rigorous proof of convergence is provided by invoking the discrete-time LaSalle's invariance principle, and we also validate our theoretical findings on quadcopters equipped with ultra-wideband ranging sensors and optical flow sensors in a GPS-less environment.