IROS 2019 论文列表

This paper presents a dynamic whole-body control method for humanoids to render agile and stable standing-up motion based on energy concepts. First, to cope with the standing-up problem with multiple contacts in hierarchical tasks, an enhanced operational-space based whole-body control (WBC) framework is proposed, which offers optimal torque resolutions guaranteeing strict dynamic consistency with inequality constraints formulated by quadratic programming. Second, agile standing-up control strategy with dynamic push and rise actions is newly developed based on the notion of the total energy. The optimal pushing wrenches at contacts are computed to obtain sufficient energy to accelerate the center-of-mass (CoM) of the robot as quick as possible, and the total energy is then controlled to attain rapid rise-up motion and to stabilize the body of the robot. Consequently, the robot can effectively and actively stand up to recover from a certain pose in which cannot be accomplished by any quasi-static motion. The proposed method is numerically experimented and validated with dynamic parameters from the real humanoid COMAN+, fulfilling different types of standing-up actions.

In autonomous navigation, a planning system reasons about other agents to plan a safe and plausible trajectory. Before planning starts, agents are typically processed with computationally intensive models for recognition, tracking, motion estimation and prediction. With limited computational resources and a large number of agents to process in real time, it becomes important to efficiently rank agents according to their impact on the decision making process. This allows spending more time processing the most important agents. We propose a system to rank agents around an autonomous vehicle (AV) in real time. We automatically generate a ranking data set by running the planner in simulation on real-world logged data, where we can afford to run more accurate and expensive models on all the agents. The causes of various planner actions are logged and used for assigning ground truth importance scores. The generated data set can be used to learn ranking models. In particular, we show the utility of combining learned features, via a convolutional neural network, with engineered features designed to capture domain knowledge. We show the benefits of various design choices experimentally. When tested on real AVs, our system demonstrates the capability of understanding complex driving situations.

Training end-to-end deep robot policies requires a lot of domain-, task-, and hardware-specific data, which is often costly to provide. In this work, we propose to tackle this issue by employing a deep neural network with a modular architecture, consisting of separate perception, policy, and trajectory parts. Each part of the system is trained fully on synthetic data or in simulation. The data is exchanged between parts of the system as low-dimensional latent representations of affordances and trajectories. The performance is then evaluated in a zero-shot transfer scenario using Franka Panda robot arm. Results demonstrate that a low-dimensional representation of scene affordances extracted from an RGB image is sufficient to successfully train manipulator policies. We also introduce a method for affordance dataset generation, which is easily generalizable to new tasks, objects and environments, and requires no manual pixel labeling.

Being motivated by ceiling inspection applications via unmanned aerial vehicles (UAVs) which require close proximity flight to surfaces, a systematic control approach enabling safe and accurate close proximity flight is proposed in this work. There are two main challenges for close proximity flights: (i) the trust characteristics varies drastically for the different distance from the ceiling which results in a complex nonlinear dynamics; (ii) the system needs to consider physical and environmental constraints to safely fly in close proximity. To address these challenges, a novel framework consisting of a constrained optimization-based force estimation and an optimization-based nonlinear controller is proposed. Experimental results illustrate that the performance of the proposed control approach can stabilize UAV down to 1 cm distance to the ceiling. Furthermore, we report that the UAV consumes up to 12.5% less power when it is operated 1 cm distance to ceiling, which is promising potential for more battery-efficient inspection flights.

This paper describes a computationally-enhanced M100 UAV platform with an onboard deep learning inference system for integrated computer vision and navigation. The system is able to autonomously find and visually identify by coat pattern individual Holstein Friesian cattle in freely moving herds. We propose an approach that utilises three deep convolutional neural network architectures running live onboard the aircraft: (1) a YOLOv2-based species detector, (2) a dual-stream deep network delivering exploratory agency, and (3) an InceptionV3-based biometric long-term recurrent convolutional network for individual animal identification. We evaluate the performance of each of the components offline, and also online via real-world field tests comprising 147 minutes of autonomous low altitude flight in a farm environment over a dispersed herd of 17 heifer dairy cows. We report error-free identification performance on this online experiment. The presented proof-of-concept system is the first of its kind. It represents a practical step towards autonomous biometric identification of individual animals from the air in open pasture environments for tag-less AI support in farming and ecology.

PiDrone is a quadrotor platform created to accompany an introductory robotics course. Students build an autonomous flying robot from scratch and learn to program it through assignments and projects. Existing educational robots do not have significant autonomous capabilities, such as high-level planning and mapping. We present a hardware and software framework for an autonomous aerial robot, in which all software for autonomy can run onboard the drone, implemented in Python. We present an Unscented Kalman Filter (UKF) for accurate state estimation. Next, we present an implementation of Monte Carlo (MC) Localization and FastSLAM for Simultaneous Localization and Mapping (SLAM). The performance of UKF, localization, and SLAM is tested and compared to ground truth, provided by a motion-capture system. Our evaluation demonstrates that our autonomous educational framework runs quickly and accurately on a Raspberry Pi in Python, making it ideal for use in educational settings.

The purpose of this study is for a robot to learn picking motions in a logistics warehouse environment. The picking operation performed by a robot often fails owing to the inclination of items placed on a shelf, as well as the minimum clearance between the products and their vinyl packaging. Therefore, we considered acquiring a specific motion trajectory by reinforcement learning. However, because numerous types of items are handled in logistics warehouses, efficient learning is required. Therefore, in this research, we propose a method to efficiently exploration for learning picking an object by determining a focus exploration area for learning based on previous results of different objects.

Letting humans and robots share a common space for collaboration is considered a consolidated practice. The trajectories followed by the robot must be safe for the human mate, especially when the robot holds dangerous tools or parts. At the same time, the productivity must be preserved, without imposing too restrictive limitations on the robot's movements. This article proposes the use of Gaussian Processes to predict the motion of an operator in a robotic cell, with the aim of controlling the robot speed and avoid collisions. An adaptive approach is proposed and the model for the human motion is persistently re-updated. The resulting approach will be demonstrated to be less conservative than previous ones, while at the same time to preserve the safety of the operator. Real experiments have been conducted on the 7 d.o.f. ABB YuMi robot.

While the mechanism of Rope-Climbing provides much flexibility, it opens up challenges to the development of the controller for Robotic Manipulator installed on Rope-Climbing robot(RCR), which is called Rope-Climbing Robot Manipulator(RCRM) here. In particular, the deformable nature of the rope results in the vibration to the manipulator and hence affects the positioning of the end effector. In this paper, a new adaptive vision-based controller is proposed for RCRM, which enables the robot to carry out the high-accuracy task under the unknown vibration from the rope. The proposed controller guarantees the performance of the robot in twofold. First, the control problem is directly formulated in the image space such that the exact spatial relationship between the moving base of the manipulator (due to the vibrating rope) and the target (e.g. the wall) is not required. Second, novel adaptation laws are developed to estimate the vibration from the rope online and are cancelled out in the robot control input to stabilize the end effector. The stability of the closed-loop system is rigorously proved with Lyapunov methods, and experimental results are presented to illustrate the performance of the proposed controller.

This paper presents a novel nonlinear disturbance rejection approach for high precision model-based control of hydraulic robots. While most disturbance rejection approaches make use of observers, we propose a novel adaptive Unscented Kalman Filter to estimate the disturbances in an unbiased minimum-variance sense. The filter is made adaptive such that there is no need to tune the covariance matrix for the disturbance estimation. Furthermore, whereas most model-based control approaches require the linearization of the system dynamics, our method is nonlinear which means that no linearization is required. Through extensive simulations as well as real hardware experiments, we demonstrate that our proposed approach can achieve high precision tracking and can be readily applied to most robotic systems even in the presence of uncertainties and external disturbances. The proposed approach is also compared to existing approaches which demonstrates its superior tracking performance.

In this paper, we tackle the problem of trajectory planning and control of a vehicle under locally varying traction limitations, in the presence of suddenly appearing obstacles. We employ concepts from adaptive model predictive control for run-time adaptation of tire force constraints that are imposed by local traction conditions. To solve the resulting optimization problem for real-time control synthesis with such time varying constraints, we propose a novel numerical scheme based on Real Time Iteration Sequential Quadratic Programming (RTI-SQP), which we call Sampling Augmented Adaptive RTI (SAA-RTI). Sampling augmentation of conventional RTI-SQP provides additional feasible candidate trajectories for warm-starting the optimization procedure. Thus, the proposed SAA-RTI algorithm enables real time constraint adaptation and reduces sensitivity to local minima. Through extensive numerical simulations we demonstrate that our method increases the vehicle's capacity to avoid accidents in scenarios with unanticipated obstacles and locally varying traction, compared to equivalent non-adaptive control schemes and traditional planning and tracking approaches.

Reduced models allow autonomous robots to cope with the complexity of planning in stochastic environments by simplifying the model and reducing its accuracy. The solution quality of a reduced model depends on its fidelity. We present 0/1 reduced model that selectively improves model fidelity in certain states by switching between using a simplified deterministic model and the full model, without significantly compromising the run time gains. We measure the reduction impact for a reduced model based on the values of the ignored outcomes and use this as a heuristic for outcome selection. Finally, we present empirical results of our approach on three different domains, including an electric vehicle charging problem using real-world data from a university campus.

This paper proposed an adaptive neural admittance control strategy for collision avoidance in human-robot collaborative tasks. In order to ensure that the robot end-effector can avoid collisions with surroundings, robot should be operated compliantly by human within a constrained task space. An impedance model and a soft saturation function are employed to generate a differentiable reference trajectory. Then, adaptive neural network control with position constraint, based on integral barrier Lyapunov function (IBLF), is designed to achieve precise tracking while guaranteeing constrained satisfaction. Utilizing Lyapunov stability principles, we prove that semi-globally uniformly bounded stability is guaranteed for all states of the closed-loop system. At last, the effectiveness of the proposed algorithm is verified on a Baxter robot experimental platform. Collisions with surroundings can be avoided in human-robot collaborative tasks.

Underwater robot interventions require a high level of safety and reliability. A major challenge to address is a robust and accurate acquisition of localization estimates, as it is a prerequisite to enable more complex tasks, e.g. floating manipulation and mapping. State-of-the-art navigation in commercial operations, such as oil& gas production (OGP), rely on costly instrumentation. These can be partially replaced or assisted by visual navigation methods, especially in deep-sea scenarios where equipment deployment has high costs and risks. Our work presents a multimodal approach that adapts state-of-the-art methods from on-land robotics, i.e., dense point cloud generation in combination with plane representation and registration, to boost underwater localization performance. A (a) two-stage navigation scheme is proposed that initially generates a coarse probabilistic map of the workspace, which is used to filter noise from computed point clouds and planes in the second stage. Furthermore, an adaptive decision-making approach is introduced that determines which perception cues to incorporate into the localization filter to optimize accuracy and computation performance. Our approach is investigated first in simulation and then validated with data from field trials in OGP monitoring and maintenance scenarios.

Object instance recognition and 3D pose estimation are important elements in robot vision technology. State-of-the-art methods improve the accuracy of both instance recognition and pose estimation using multitask learning. These methods use unified balancing parameters to integrate the loss of each task, which means task difficulties are the same for all objects. However, the method we propose can adjust the balancing parameters for each object. This idea is based on the assumption that task difficulties are different for each object, since the distinctiveness of object instances and poses depends on their appearance and shape. Our method sequentially estimates task difficulties for CNN based on the amount of loss change and calculates balancing parameters for each object. Our experiments show that our method improves the accuracy of both object instance recognition and pose estimation compared with state-of-the-art methods using the common LineMOD dataset.

We propose a deep reinforcement learning (DRL) methodology for the tracking, obstacle avoidance, and formation control of nonholonomic robots. By separating vision-based control into a perception module and a controller module, we can train a DRL agent without sophisticated physics or 3D modeling. In addition, the modular framework averts daunting retrains of an image-to-action end-to-end neural network, and provides flexibility in transferring the controller to different robots. First, we train a convolutional neural network (CNN) to accurately localize in an indoor setting with dynamic foreground/background. Then, we design a new DRL algorithm named Momentum Policy Gradient (MPG) for continuous control tasks and prove its convergence. We also show that MPG is robust at tracking varying leader movements and can naturally be extended to problems of formation control. Leveraging reward shaping, features such as collision and obstacle avoidance can be easily integrated into a DRL controller.

Coverage path planning of a known environment sees a variety of applications, including cleaning, surveillance, agriculture and 3D printing. Most approaches employ hard-coded heuristics or other application-specific requirements, making them hard to extend to other problem scenarios or “configurations,” such as different motion strategies or robot size. This work presents a unifying, general, and adaptive framework, called adaptive deep path (AD Path), for coverage path planning problems under a variety of configurations. It can improve path efficiency with respect to both path length and number of turns, and can flexibly accommodate different problem configuration options. We evaluate AD Path against a state-of-the-art baseline in four complex environments with different configurations. We show that our approach can produce efficient paths; our experimental results show that AD Path can reduce the path length by 21.8% and the number turns by 38.6% on average compared with the baseline.

In robot-assisted rehabilitation, assist-as-needed (AAN) controllers have been proposed to promote subjects’ active participation, which is thought to lead to better training outcomes. Most of these AAN controllers require a patient-specific manual tuning of the parameters defining the underlying force-field, which typically results in a tedious and time-consuming process. In this paper, we propose a reinforcement-learning-based impedance controller that actively reshapes the stiffness of the force-field to the subject’s performance, while providing assistance only when needed. This adaptability is made possible by correlating the subject’s most recent performance to the ultimate control objective in real-time. In addition, the proposed controller is built upon action dependent heuristic dynamic programming using the actor-critic structure, and therefore does not require prior knowledge of the system model. The controller is experimentally validated with healthy subjects through a simulated ankle mobilization training session using a powered ankle-foot orthosis.

Deep neural networks (DNNs) are being incorporated into various autonomous systems like self-driving cars and robots. However, there is a rising concern about the robustness of these systems because of their susceptibility to adversarial attacks on DNNs. Past research has established that DNNs used for classification and object detection are prone to attacks causing targeted misclassification. In this paper, we show the effectiveness of an adversarial dynamic attack on an end-to-end trained DNN controlling an autonomous vehicle. We launch the attack by installing a billboard on the roadside and displaying videos to approaching vehicles to cause the DNN controller in the vehicle to generate steering commands that cause, for example, unintended lane changes or motion off the road causing accidents. The billboard has an integrated camera estimating the pose of the on-coming vehicle. The approach enables dynamic adversarial perturbation that adapts to the relative pose of the vehicle and uses the dynamics of the vehicle to steer it along adversary-chosen trajectories while being robust to variations in view, lighting, and weather. We demonstrate the effectiveness of the attack on a recently published off-the-shelf end-to-end learning-based autonomous navigation system in a high-fidelity simulator, CARLA (CAR Learning to Act). The proposed approach may also be applied to other systems driven by an end-to-end trained network.