IROS 2016 Papers

We propose a novel method that arbitrates the control between the human and the robot actors in a teaching-by-demonstration setting to form synergy between the two and facilitate effective skill synthesis on the robot. We employed the human-in-the-loop teaching paradigm to teleoperate and demonstrate a complex task execution to the robot in real-time. As the human guides the robot to perform the task, the robot obtains the skill online during the demonstration. To encode the robotic skill we employed Locally Weighted Regression that fits local models to specific state region of the task based on the human demonstration. If the robot is in the state region where no local models exist, the control over the robotic mechanism is given to the human to perform the teaching. When local models are gradually obtained in that region, the control is given to the robot so that the human can examine its performance already during the demonstration stage, and take actions accordingly. This enables a co-adaptation between the agents and contributes to a faster and more efficient teaching. As a proof-of-concept, we realised the proposed robot teaching system on a haptic robot with the task of generation of a desired vertical force on a horizontal plane with unknown stiffness properties.

Modeling of physical human-robot collaborations is generally a challenging problem due to the unpredictive nature of human behavior. To address this issue, we present a data-efficient reinforcement learning framework which enables a robot to learn how to collaborate with a human partner. The robot learns the task from its own sensorimotor experiences in an unsupervised manner.

An important aspect to be taken care of while designing assistive robots for mobility is that they need to operate among humans. Thus understanding human spatial social conventions and incorporating them in the assistive solutions, is important. In this paper, we introduce a semi-autonomous framework for assistive wheelchair navigation in human environments, which is driven by the intention of the wheelchair user. Safe and socially compliant motion provided by a user intention driven local motion planner is fused with user teleoperation in order to create such a system. Taking into account the fact that the user is the primary controller, our proposed system aims to provide progressive assistance whenever the user is in danger of collision or at risk of disturbance to other humans. We also thus propose generalized formulations for estimating user intentions and for sharing control within the context of wheelchair mobility assistance, that is adaptable in order to be deployed in real world systems. We then evaluate the proposed framework in simulation in order to obtain a quantitative analysis. We also provide experimental evidence using an off-the-shelf robotized wheelchair equipped with a single 2D laser scanner.

To accelerate the developmental process of robots interacting with humans in the real world, we propose the simultaneous introduction of two competing robots into the test field. One robot moves autonomously using a preprogrammed behavior set, while the other is remotely controlled by a human operator. The operator attempts to improve the robot by exploring new behavioral elements. Concurrently, the preprogrammed behavior set is tested by the first robot. The main concept is that by allowing the two robots compete against each other, we aim to accelerate the development process. By applying this methodology, we developed an educational robot for children. Herein we report the functioning of this methodology and how behavioral elements were explored and improved through field development.

This paper introduces a new robotic gripper for flat surfaces based on a novel electrostatic/gecko-like adhesive. This unique gripping solution overcomes the shortcomings of vacuum grippers for part-handling by eliminating the need for a compressed air system and offering more rapid actuation, thus achieving significant potential cost savings and throughput improvements in manufacturing processes. Results demonstrate the gripper's performance on a variety of both smooth and rough surfaces, including fabrics, as well as show that the gripper is able to successfully pick up and release glass and carbon fiber sheets for over 100 cycles.

We propose a topology-based Bayesian visual navigation framework, with which we represent the world environment as a collection of scenes. The proposed topological map consists of nodes represented as a bag of visual line words and edges represented as both an adjacency list and relative motion information to perform the transition between topological nodes. Our proposed Bayesian localization framework uses two measurement models: a visual line word-based place model and a path-matching model. To enable a mobile robot to reach to a desired destination, a coastal path is planned in such a way that maximizes the possibility of following the reference path defined as a sequence of scenes encountered when building the topological map, while avoiding possible collisions by leveraging a local grid map constructed in real time from Kinect depth information. To show the validity of our proposed framework, we provide several extensive experimental results in several indoor environments under different conditions, such as illumination changes and visual occlusions.

Omnidirectional mobile robots with four Mecanum wheels are used in various homes, military, nuclear power plant, industrial, hospital and space applications. A lot of research has been done using three universal wheels but trajectory control for four Mecanum wheeled mobile robot (FMWMR) in presence of uncertainties still needs attention. Thus, to obtain smooth motion of mobile robots, with chattering free control input, in presence of uncertainties and external force disturbances, a robust and adaptive control is necessary. In view of these aspects, this paper extends the use of adaptive sliding mode controller for trajectory tracking of FMWMR. The effectiveness of proposed controller is verified using two case problems. Simulation results are presented for the verification of proposed controller for FMWMR. Further, experiments are conducted using position and orientation sensor to show the performance of the controller in real world environment. Simulation and experimental results revealed that FMWMR is capable of tracking any type of trajectories.

Purpose of this study is to propose a new robotic system equipped with a steerable catheter for percutaneous coronary intervention. Key design issues of the system are less X-ray exposure, easy sterilization, and convenient user interface.

This paper considers the problem of controlling a formation of quadrotor UAVs equipped with onboard cameras able to measure relative bearings in their local body frames w.r.t. neighboring UAVs. The control goal is twofold: (i) steering the agent group towards a formation defined in terms of desired bearings, and (ii) actuating the group motions in the ‘null-space’ of the current bearing formation. The proposed control strategy relies on an extension of the rigidity theory to the case of directed bearing frameworks in ℝ3×S1. This extension allows to devise a decentralized bearing controller which, unlike most of the present literature, does not need presence of a common reference frame or of reciprocal bearing measurements for the agents. Simulation and experimental results are then presented for illustrating and validating the approach.

This paper presents a deployable boom which combines a rigid telescopic frame and a flexible tape spring. The front end of the spring is fixed on the rear end of the innermost segment of the frame. The spring spreads and rolls up inside the frame to drive the segments to move one by one to realize the boom deployment and retraction. The driving forces needed to deploy and retract the frame are modeled and simulated. The feasibility of the frame driving by only one spring is also studied. A 1.6 kg prototype system with 2.1 m total deployment length is implemented. Experimental results show the maximum driving forces for deployment and retraction of the frame are about 9.1 N and 6.8 N respectively. The boom is able to deploy in 76 s with energy consumption of 315 J. The boom can resist at least 15 N force axially and 31.5 N·m bending moment when the forces are acted on its front end. Advantages of this kind of boom enable it to be applied for instruments deployment, walking and sampling assists, and robotic arms design in space exploration.

Many robot learning algorithms depend on a model of the robot's forward dynamics for simulating potential trajectories and ultimately learning a required task. In this paper, we present a data-driven reservoir computing approach and apply it for learning forward dynamics models. Our proposed machine learning algorithm exploits the concepts of dynamic reservoir, self-organized learning and Bayesian inference. We have evaluated our approach on datasets gathered from two industrial robotic manipulators and compared it on both step-by-step and multi-step trajectory prediction scenarios with state-of-the-art algorithms. The evaluation considers the algorithms' convergence and prediction performance on joint and operational space for varying prediction horizons, as well as computational time. Results show that the proposed algorithm performs better than the state-of-the-art, converges fast and can achieve accurate predictions over longer horizons, which makes it a reliable, data-efficient approach for learning forward models.

Reactive stepping is an important utility to regain balance when bipedal walking motions are disturbed. This paper sheds light on the reasons for humanoid robots to fall down. It presents a method to calculate modifications of predefined foot placements with the objective to minimize deviations of the Zero Moment Point from a reference without interrupting the walk. The calculation is in closed-form, and is embedded into a well-evaluated preview controller with observer based on the 3D Linear Inverted Pendulum Mode (3D-LIPM). Experiments in simulation and on a physical robot prove the benefit of the proposed system.

We present a novel method for guiding a large-scale swarm of autonomous agents into a desired formation shape in a distributed and scalable manner. Our Probabilistic Swarm Guidance using Inhomogeneous Markov Chains (PSG-IMC) algorithm adopts an Eulerian framework, where the physical space is partitioned into bins and the swarm's density distribution over each bin is controlled. Each agent determines its bin transition probabilities using a time-inhomogeneous Markov chain. These time-varying Markov matrices are constructed by each agent in real-time using the feedback from the current swarm distribution, which is estimated in a distributed manner. The PSG-IMC algorithm minimizes the expected cost of the transitions per time instant, required to achieve and maintain the desired formation shape, even when agents are added to or removed from the swarm. The algorithm scales well with a large number of agents and complex formation shapes. We demonstrate the effectiveness of this proposed swarm guidance algorithm by using results of numerical simulations and hardware experiments with multiple quadrotors.

Robotic assistants have the potential to greatly improve our quality of life by supporting us in our daily activities. A service robot acting autonomously in an indoor environment is faced with very complex tasks. Consider the problem of pouring a liquid into a cup, the robot should first determine if the cup is empty or partially filled. RGB-D cameras provide noisy depth measurements which depend on the opaqueness and refraction index of the liquid. In this paper, we present a novel probabilistic approach for estimating the fill-level of a liquid in a cup using an RGB-D camera. Our approach does not make any assumptions about the properties of the liquid like its opaqueness or its refraction index. We develop a probabilistic model using features extracted from RGB and depth data. Our experiments demonstrate the robustness of our method and an improvement over the state of the art.

The objective characterization of human motion is required in a variety of fields including competitive sports, rehabilitation and the detection of motor deficits. Nowadays, typically human experts evaluate the motor behavior. These evaluations are based on their individual experience which leads to a low inter- and intra-expert reliability. Standardized tests improve on the reliability but are still prone to subjective ratings and require human expert knowledge. This paper presents a novel method to characterize the motor state of Parkinson patients using full body motion capturing data based on a combination of multiple metrics. Our approach merges various metrics with a Random Forest and uses a probabilistic formulation to compute a one-dimensional measure for the performed motion. We present an application of our approach to the problem of relating subject motion to different classes like healthy subjects and Parkinson disease patients with deep brain stimulation switched on or off. In the experimental session we show that our measure leads to high classification rates and high entropy values for real-world data. Besides, we show that our method discriminates between Parkinson's subjects (with and without stimulation) and healthy persons as good as the Unified Parkinson's Disease Rating Scale (UPDRS).

This paper reviews the motivation, development, and use of underwater robotic vehicles designed for use in ice-covered waters, with special attention paid to the navigation systems employed for under-ice deployments. The scientific need for routine access under fixed and moving ice by underwater robotic vehicles is briefly reviewed. The challenges of under-ice vehicle navigation are summarized. The paper then reviews all known under-ice robotic vehicles and their associated navigation systems, categorizing them by vehicle type (tethered, untethered, hybrid, and glider) and by the type of ice they were designed for (fixed land-fast sea ice and moving sea ice).

New work in robotics targets the development of controllable agile motions such as leaping. In this work, we examine animal and robotic systems on the metric of jumping agility and find that animals can outperform the most agile robots by a factor of two. These specially adapted animals use a jumping strategy we term power modulation to generate more peak power for jumping than otherwise possible. A novel eight-bar revolute mechanism designed with a new linkage synthesis approach encodes the properties for power modulation as well as constraints which assure rotation-free jumping motion. We fabricate an 85 gram prototype and demonstrate that it can perform a range of jumps while constrained by a linear slide. The prototype can deliver 3.63 times more peak jumping power than the maximum its motor can produce. A simulation matched to the physical parameters of the prototype predicts that the robot can attain an agility exceeding that of the most agile animals if the actuator power is increased to 15W.

In this paper, we propose a novel planar biped model which achieves asymptotically stable walking, based on ankle actuation and the virtual pendulum concept. Simple models for walking have provided useful insights in understanding fundamental principles of human locomotion as well as developing controllers for biped robots. Existing simplest walking models include a point mass with either rigid legs or compliant legs. Both models are able to describe different gaits such as walking and running, but are not able to address posture stabilization, which is another important issue in bipedalism. Recently, the virtual pendulum (VP) concept was proposed as an intuitive posture stabilization strategy and successfully demonstrated on the compliant leg scheme. However, the model does not address inherent peripheral mechanics of walking, i.e., ankle actuation and collisional energy loss from foot-ground interaction. The model proposed in this paper consists of a rigid trunk, rigid legs and ankle actuation. The energy dissipation due to collision and compensation by ankle actuation play an essential role for asymptotically stable walking, while the trunk posture is stabilized based on the VP concept. Nonlinear dynamic simulation verifies that the model constructs a stable limit cycle with a small angular oscillation of trunk.