机器人相关学术速递[11.10]

cs.RO机器人相关，共计11篇

【1】 A Differentiable Recipe for Learning Visual Non-Prehensile Planar Manipulation 标题：一种学习视觉非抓取平面操作的微分配方 链接：https://arxiv.org/abs/2111.05318

作者：Bernardo Aceituno,Alberto Rodriguez,Shubham Tulsiani,Abhinav Gupta,Mustafa Mukadam 机构：Massachusetts Institute of Technology,Facebook AI Research 备注：Presented at CORL 2021 摘要：通过视频指定任务是获得新颖和通用机器人技能的一种强大技术。然而，对力学和灵巧交互的推理可能会使衡量学习接触丰富的操作具有挑战性。在这项工作中，我们将重点放在视觉非可理解的平面操纵问题上：给定一个平面运动对象的视频，找到再现相同对象运动的接触感知机器人动作。我们提出了一种新的结构，可微操作学习（\ours），它通过利用可微优化和基于有限差分的模拟，将视频解码神经模型与接触力学的先验知识相结合。通过大量的模拟实验，我们研究了传统的基于模型的技术和现代深度学习方法之间的相互作用。我们发现，我们的模块化和完全可微的体系结构在看不见的对象和运动上比只学习的方法表现得更好\网址{https://github.com/baceituno/dlm}. 摘要：Specifying tasks with videos is a powerful technique towards acquiring novel and general robot skills. However, reasoning over mechanics and dexterous interactions can make it challenging to scale learning contact-rich manipulation. In this work, we focus on the problem of visual non-prehensile planar manipulation: given a video of an object in planar motion, find contact-aware robot actions that reproduce the same object motion. We propose a novel architecture, Differentiable Learning for Manipulation (\ours), that combines video decoding neural models with priors from contact mechanics by leveraging differentiable optimization and finite difference based simulation. Through extensive simulated experiments, we investigate the interplay between traditional model-based techniques and modern deep learning approaches. We find that our modular and fully differentiable architecture performs better than learning-only methods on unseen objects and motions. \url{https://github.com/baceituno/dlm}.

【2】 Designing and Analyzing the PID and Fuzzy Control System for an Inverted Pendulum 标题：倒立摆的PID和模糊控制系统设计与分析 链接：https://arxiv.org/abs/2111.05309

作者：Armin Masoumian,Pezhman kazemi,Mohammad Chehreghani Montazer,Hatem A. Rashwan,Domenec Puig Valls 机构：Department of Computer Engineering and Mathematics, Universitat Rovira I Virgili, Tarragona, Spain, Departament of Chemical Engineering, Mechatronics Department, University of Debrecen, Debrecen, Hungary 备注：5 pages, Accepted for The 6th International Conference on Mechatronics and Robotics Engineering (ICMRE 2020) 摘要：倒立摆是一个非线性不平衡系统，需要使用电机来控制，以实现稳定和平衡。倒立摆由乐高制作，使用乐高Mindstorm NXT，这是一种可编程机器人，能够完成许多不同的功能。本文提出了倒立摆的初步设计，并研究了与乐高Mindstorm NXT兼容的不同传感器的性能。此外，还研究了计算机视觉实现维护系统所需稳定性的能力。倒立摆是一种传统的小车，可以使用模糊逻辑控制器进行控制，该控制器为小车的移动提供自调整PID控制。在MATLAB和Simulink中对模糊逻辑和PID进行了仿真，并在LabVIEW软件中开发了机器人程序。 摘要：The inverted pendulum is a non-linear unbalanced system that needs to be controlled using motors to achieve stability and equilibrium. The inverted pendulum is constructed with Lego and using the Lego Mindstorm NXT, which is a programmable robot capable of completing many different functions. In this paper, an initial design of the inverted pendulum is proposed and the performance of different sensors, which are compatible with the Lego Mindstorm NXT was studied. Furthermore, the ability of computer vision to achieve the stability required to maintain the system is also investigated. The inverted pendulum is a conventional cart that can be controlled using a Fuzzy Logic controller that produces a self-tuning PID control for the cart to move on. The fuzzy logic and PID are simulated in MATLAB and Simulink, and the program for the robot is developed in the LabVIEW software.

【3】 Using The Feedback of Dynamic Active-Pixel Vision Sensor (Davis) to Prevent Slip in Real Time 标题：利用动态主动像素视觉传感器(Davis)的反馈进行实时防滑 链接：https://arxiv.org/abs/2111.05308

作者：Armin Masoumian,Pezhman kazemi,Mohammad Chehreghani Montazer,Hatem A. Rashwan,Domenec Puig Valls 机构：Department of Computer Engineering and Mathematics, Universitat Rovira I Virgili, Tarragona, Spain, Departament of Chemical Engineering, Mechatronics Department, University of Debrecen, Debrecen, Hungary 备注：5 pages, Accepted for The 6th International Conference on Mechatronics and Robotics Engineering (ICMRE 2020) 摘要：本文的目的是描述一种在实时反馈中检测滑动和接触力的方法。在这种新的方法中，戴维斯相机由于其快速的处理速度和高分辨率被用作视觉触觉传感器。在四个不同形状、大小、重量和材料的物体上进行了200次实验，以比较巴克斯特机器人夹持器的精度和响应，从而避免滑动。通过使用力敏电阻器（FSR402）验证了先进的方法。戴维斯摄像机捕捉到的事件通过特定算法进行处理，以向巴克斯特机器人提供反馈，帮助其检测滑倒。 摘要：The objective of this paper is to describe an approach to detect the slip and contact force in real-time feedback. In this novel approach, the DAVIS camera is used as a vision tactile sensor due to its fast process speed and high resolution. Two hundred experiments were performed on four objects with different shapes, sizes, weights, and materials to compare the accuracy and response of the Baxter robot grippers to avoid slipping. The advanced approach is validated by using a force-sensitive resistor (FSR402). The events captured with the DAVIS camera are processed with specific algorithms to provide feedback to the Baxter robot aiding it to detect the slip.

【4】 Learning Perceptual Concepts by Bootstrapping from Human Queries 标题：从人类查询中通过自举学习知觉概念 链接：https://arxiv.org/abs/2111.05251

作者：Andreea Bobu,Chris Paxton,Wei Yang,Balakumar Sundaralingam,Yu-Wei Chao,Maya Cakmak,Dieter Fox 机构：com 3 University of Washington mcakmak 备注：7 pages, 7 figures 摘要：机器人需要能够从用户那里学习概念，以便使其能力适应每个用户的独特任务。但当机器人在高维输入（如图像或点云）上操作时，这是不切实际的：机器人需要不切实际的人力来学习新概念。为了应对这一挑战，我们提出了一种新的方法，即机器人学习概念的低维变体，并使用它生成更大的数据集，以便在高维空间学习概念。这使得它可以利用只有在训练时才可访问的语义上有意义的特权信息，如对象姿势和边界框，从而允许更丰富的人机交互来加速学习。我们通过学习描述对象状态或多对象关系的介词概念来评估我们的方法，如上图、近图或对齐图，它们是用户指定机器人任务目标和执行约束的关键。使用模拟人，我们表明，与直接在高维空间学习概念相比，我们的方法提高了样本复杂性。我们还演示了所学概念在7自由度Franka Panda机器人运动规划任务中的实用性。 摘要：Robots need to be able to learn concepts from their users in order to adapt their capabilities to each user's unique task. But when the robot operates on high-dimensional inputs, like images or point clouds, this is impractical: the robot needs an unrealistic amount of human effort to learn the new concept. To address this challenge, we propose a new approach whereby the robot learns a low-dimensional variant of the concept and uses it to generate a larger data set for learning the concept in the high-dimensional space. This lets it take advantage of semantically meaningful privileged information only accessible at training time, like object poses and bounding boxes, that allows for richer human interaction to speed up learning. We evaluate our approach by learning prepositional concepts that describe object state or multi-object relationships, like above, near, or aligned, which are key to user specification of task goals and execution constraints for robots. Using a simulated human, we show that our approach improves sample complexity when compared to learning concepts directly in the high-dimensional space. We also demonstrate the utility of the learned concepts in motion planning tasks on a 7-DoF Franka Panda robot.

【5】 Robot control for simultaneous impact tasks via QP based reference spreading 标题：基于QP的参考扩散实现同时冲击任务的机器人控制 链接：https://arxiv.org/abs/2111.05211

作者：Jari J. van Steen,Nathan van de Wouw,Alessandro Saccon 备注：Submitted to IEEE for the ACC 2022 conference 摘要：为了进一步利用机器人操作中的碰撞，提出了一种控制框架，该框架直接解决了机器人跟踪控制带来的挑战，机器人的任务是执行与多个接触点相关的名义上同时碰撞。为此，我们扩展了参考扩展框架，该框架使用了与刚性碰撞图一致的扩展碰撞前后参考，在非弹性同时碰撞的假设下确定。在实践中，机器人在碰撞时刻不会精确地停留在参考点上；因此，通常会在不同接触点发生一系列冲击。我们的新方法通过引入额外的中间控制模式，扩展了这种情况下的参考传播。在此模式下，扭矩指令仍然基于前碰撞参考，目标是达到目标接触状态，但速度反馈被禁用，因为这可能由于速度快速变化而造成潜在危害。着眼于实际实现，该方法采用了QP控制框架，并在刚性机器人模型和具有柔性关节的真实机器人模型上进行了数值模拟验证。 摘要：With the aim of further enabling the exploitation of impacts in robotic manipulation, a control framework is presented that directly tackles the challenges posed by tracking control of robotic manipulators that are tasked to perform nominally simultaneous impacts associated to multiple contact points. To this end, we extend the framework of reference spreading, which uses an extended ante- and post-impact reference coherent with a rigid impact map, determined under the assumption of an inelastic simultaneous impact. In practice, the robot will not reside exactly on the reference at the impact moment; as a result a sequence of impacts at the different contact points will typically occur. Our new approach extends reference spreading in this context via the introduction of an additional intermediate control mode. In this mode, a torque command is still based on the ante-impact reference with the goal of reaching the target contact state, but velocity feedback is disabled as this can be potentially harmful due to rapid velocity changes. With an eye towards real implementation, the approach is formulated using a QP control framework and is validated using numerical simulations both on a rigid robot model and on a realistic robot model with flexible joints.

【6】 Footstep Adjustment for Biped Push Recovery on Slippery Surfaces 标题：滑面上Biped推力恢复的足迹调整 链接：https://arxiv.org/abs/2111.05203

作者：Erfan Ghorbani,Hossein Karimpour,Venus Pasandi,Mehdi Keshmiri 机构：ir 1Department of Mechanical Engineering, Isfahan Universityof Technology, Iran 2Department of Mechanical Engineering, University of Isfahan 备注：for associated simulation video, see this https URL 摘要：尽管对两足动物的运动稳定性进行了广泛的研究，但它们仍然缺乏在光滑表面上应对干扰的能力。在本文中，一种新的控制器，用于稳定两足动物在其矢状面上的运动，并考虑到表面摩擦的限制。通过考虑稳定趋势中表面的物理限制，实现了更高级的可靠性，提供了更高的功能，如低摩擦表面上的推力恢复，并防止稳定器过度反应。基于离散事件的策略包括修改每个足迹开始时的步长和时间周期，以重新建立稳定性必要条件，同时考虑表面摩擦限制作为防止滑动的约束。在面对外部干扰时，调整脚步以防止滑动被认为是保持稳定的一种新策略，与人类的反应非常相似。所开发的方法包括利用初等数学运算获得控制输入的粗糙闭式解，允许在收敛性和计算成本之间达成平衡，这非常适合于实时操作，即使使用适度的计算硬件。几个数值模拟，包括推力恢复和低摩擦表面上不同栅极之间的切换，验证了该控制器的有效性。与人类步态经验相关，研究结果还揭示了一些有利于稳定性的物理方面，以及在不同条件下切换步态以降低跌倒风险的事实。 摘要：Despite extensive studies on motion stabilization of bipeds, they still suffer from the lack of disturbance coping capability on slippery surfaces. In this paper, a novel controller for stabilizing a bipedal motion in its sagittal plane is developed with regard to the surface friction limitations. By taking into account the physical limitation of the surface in the stabilization trend, a more advanced level of reliability is achieved that provides higher functionalities such as push recovery on low-friction surfaces and prevents the stabilizer from overreacting. The discrete event-based strategy consists of modifying the step length and time period at the beginning of each footstep in order to reestablish stability necessary conditions while taking into account the surface friction limitation as a constraint to prevent slippage. Adjusting footsteps to prevent slippage in confronting external disturbances is perceived as a novel strategy for keeping stability, quite similar to human reaction. The developed methodology consists of rough closed-form solutions utilizing elementary math operations for obtaining the control inputs, allowing to reach a balance between convergence and computational cost, which is quite suitable for real-time operations even with modest computational hardware. Several numerical simulations, including push recovery and switching between different gates on low-friction surfaces, are performed to demonstrate the effectiveness of the proposed controller. In correlation with human-gait experience, the results also reveal some physical aspects favoring stability and the fact of switching between gaits to reduce the risk of falling in confronting different conditions.

【7】 Analyzing and Improving Fault Tolerance of Learning-Based Navigation Systems 标题：基于学习的导航系统容错性分析与改进 链接：https://arxiv.org/abs/2111.04957

作者：Zishen Wan,Aqeel Anwar,Yu-Shun Hsiao,Tianyu Jia,Vijay Janapa Reddi,Arijit Raychowdhury 机构：Georgia Institute of Technology, Atlanta, GA, Harvard University, Cambridge, MA 备注：Accepted in 58th ACM/IEEE Design Automation Conference (DAC), 2021 摘要：基于学习的导航系统广泛应用于机器人、无人驾驶车辆和无人机等自主应用领域。专门的硬件加速器已被提议用于此类导航任务的高性能和能源效率。然而，硬件系统中的瞬时和永久性故障正在增加，可能会严重破坏任务的安全性。同时，传统的基于冗余的保护方法难以部署在资源受限的边缘应用程序上。在本文中，我们从RL训练和推理两个方面对导航系统的算法、故障模型和数据类型进行了实验评估。我们进一步提出了两种有效的故障缓解技术，在基于学习的导航系统中实现了2倍的成功率和39%的飞行质量改进。 摘要：Learning-based navigation systems are widely used in autonomous applications, such as robotics, unmanned vehicles and drones. Specialized hardware accelerators have been proposed for high-performance and energy-efficiency for such navigational tasks. However, transient and permanent faults are increasing in hardware systems and can catastrophically violate tasks safety. Meanwhile, traditional redundancy-based protection methods are challenging to deploy on resource-constrained edge applications. In this paper, we experimentally evaluate the resilience of navigation systems with respect to algorithms, fault models and data types from both RL training and inference. We further propose two efficient fault mitigation techniques that achieve 2x success rate and 39% quality-of-flight improvement in learning-based navigation systems.

【8】 Safe Policy Optimization with Local Generalized Linear Function Approximations 标题：基于局部广义线性函数逼近的安全策略优化 链接：https://arxiv.org/abs/2111.04894

作者：Akifumi Wachi,Yunyue Wei,Yanan Sui 机构：IBM Research, Tsinghua University 备注：18 pages, 6 figures, Accepted to NeurIPS-21 摘要：安全探索是在安全关键系统中应用强化学习（RL）的关键。现有的安全勘探方法是在规律性假设下保证安全的，难以应用于大规模的实际问题。我们提出了一种新的算法SPO-LF，该算法在学习传感器获得的局部可用特征和使用广义线性函数近似的环境奖励/安全之间的关系的同时优化代理的策略。我们为其安全性和最优性提供了理论保证。我们的实验表明，我们的算法1）在样本复杂度和计算成本方面更有效，2）比以前的安全RL方法更适用于大规模问题，并且3）与现有的具有安全约束的高级深度RL方法相比，具有相当的样本效率和安全性。 摘要：Safe exploration is a key to applying reinforcement learning (RL) in safety-critical systems. Existing safe exploration methods guaranteed safety under the assumption of regularity, and it has been difficult to apply them to large-scale real problems. We propose a novel algorithm, SPO-LF, that optimizes an agent's policy while learning the relation between a locally available feature obtained by sensors and environmental reward/safety using generalized linear function approximations. We provide theoretical guarantees on its safety and optimality. We experimentally show that our algorithm is 1) more efficient in terms of sample complexity and computational cost and 2) more applicable to large-scale problems than previous safe RL methods with theoretical guarantees, and 3) comparably sample-efficient and safer compared with existing advanced deep RL methods with safety constraints.

【9】 LiMoSeg: Real-time Bird's Eye View based LiDAR Motion Segmentation 标题：LiMoSeg：基于实时鸟瞰的LiDAR运动分割 链接：https://arxiv.org/abs/2111.04875

作者：Sambit Mohapatra,Mona Hodaei,Senthil Yogamani,Stefan Milz,Patrick Maeder,Heinrich Gotzig,Martin Simon,Hazem Rashed 机构：Valeo, Germany, Valeo, Ireland, Spleenlab.ai, Germany, TU Ilmenau, Germany 摘要：运动目标检测与分割是自动驾驶管道中的一项重要任务。检测和隔离车辆周围的静态和移动部件在路径规划和定位任务中尤为重要。提出了一种新的实时光探测和测距（LiDAR）数据运动分割体系结构。我们使用二维鸟瞰图（BEV）表示的两次连续激光雷达数据扫描来执行静态或移动的像素分类。此外，我们提出了一种新的数据增强技术，以减少静态和移动对象之间的显著类不平衡。我们通过剪切和粘贴静态车辆来人工合成移动对象来实现这一点。我们在一个常用的汽车嵌入式平台，即Nvidia Jetson Xavier上演示了8毫秒的低延迟。据我们所知，这是第一个直接在LiDAR BEV空间中执行运动分割的工作。我们提供了具有挑战性的SemanticKITTI数据集的定量结果，定性结果见https://youtu.be/2aJ-cL8b0LI. 摘要：Moving object detection and segmentation is an essential task in the Autonomous Driving pipeline. Detecting and isolating static and moving components of a vehicle's surroundings are particularly crucial in path planning and localization tasks. This paper proposes a novel real-time architecture for motion segmentation of Light Detection and Ranging (LiDAR) data. We use two successive scans of LiDAR data in 2D Bird's Eye View (BEV) representation to perform pixel-wise classification as static or moving. Furthermore, we propose a novel data augmentation technique to reduce the significant class imbalance between static and moving objects. We achieve this by artificially synthesizing moving objects by cutting and pasting static vehicles. We demonstrate a low latency of 8 ms on a commonly used automotive embedded platform, namely Nvidia Jetson Xavier. To the best of our knowledge, this is the first work directly performing motion segmentation in LiDAR BEV space. We provide quantitative results on the challenging SemanticKITTI dataset, and qualitative results are provided in https://youtu.be/2aJ-cL8b0LI.

【10】 Learned Dynamics of Electrothermally-Actuated Soft Robot Limbs Using LSTM Neural Networks 标题：基于LSTM神经网络的电热驱动软机器人肢体学习动力学 链接：https://arxiv.org/abs/2111.04851

作者：Rohan K. Mehta,Andrew P. Sabelhaus,Carmel Majidi 机构： Carnegie Mellon University 备注：6 pages, 7 figures 摘要：由于热滞后和机械滞后以及机器人操作过程中可能出现的复杂物理交互作用，使用电热致动器对软机器人肢体的动力学建模通常具有挑战性。本文提出了一种基于长短时记忆（LSTM）的神经网络来解决执行器建模中的这些挑战。一个由一对形状记忆合金（SMA）线圈驱动并包含嵌入式温度和角度偏转传感器的平面软肢用作测试平台。该机器人的数据用于训练LSTM神经网络，使用不同的传感器数据组合，对单向（一个SMA）和双向（两个SMA）运动进行建模。开环展示结果表明，学习的模型能够预测超长开环时间尺度（10分钟）内的运动，且漂移很小。即使仅使用致动器的脉宽调制输入进行学习，预测误差也取决于软挠度传感器的精度。这些LSTM模型可在现场使用，无需大量传感，有助于将软电热驱动机器人带入实际应用。 摘要：Modeling the dynamics of soft robot limbs with electrothermal actuators is generally challenging due to thermal and mechanical hysteresis and the complex physical interactions that can arise during robot operation. This article proposes a neural network based on long short-term memory (LSTM) to address these challenges in actuator modeling. A planar soft limb, actuated by a pair of shape memory alloy (SMA) coils and containing embedded sensors for temperature and angular deflection, is used as a test platform. Data from this robot are used to train LSTM neural networks, using different combinations of sensor data, to model both unidirectional (one SMA) and bidirectional (both SMAs) motion. Open-loop rollout results show that the learned model is able to predict motions over extraordinarily long open-loop timescales (10 minutes) with little drift. Prediction errors are on the order of the soft deflection sensor's accuracy, even when using only the actuator's pulse width modulation inputs for learning. These LSTM models can be used in-situ, without extensive sensing, helping to bring soft electrothermally-actuated robots into practical application.

【11】 Planar Robot Casting with Real2Sim2Real Self-Supervised Learning 标题：基于Real2Sim2Real自监督学习的平面机器人铸造 链接：https://arxiv.org/abs/2111.04814

作者：Vincent Lim,Huang Huang,Lawrence Yunliang Chen,Jonathan Wang,Jeffrey Ichnowski,Daniel Seita,Michael Laskey,Ken Goldberg 机构： 1AUTOLAB at the University of California, 2Toyota Research Institute 摘要：使用单个参数化动态动作操纵可变形对象对于飞钓、放样毛毯和玩沙狐球等任务非常有用。这些任务以期望的最终状态作为输入，并输出一个参数化的开环动态机器人动作，该动作产生朝向最终状态的轨迹。这对于包含摩擦的复杂动力学的长视距轨迹尤其具有挑战性。本文探讨了平面机器人浇铸（PRC）的任务：机器人手腕的一个平面运动握住电缆的一端，导致另一端在平面上滑向所需目标。PRC允许电缆到达机器人工作空间以外的点，并应用于家庭、仓库和工厂的电缆管理。为了有效地学习给定电缆的PRC策略，我们提出了Real2Sim2Real，一个自动收集物理轨迹示例以使用差分进化调整动力学模拟器参数的自我监督框架，生成许多模拟示例，然后使用模拟数据和物理数据的加权组合学习策略。我们使用三个模拟器（Isaac Gym segmented、Isaac Gym hybrid和PyBullet）、两个函数近似器、高斯过程和神经网络（NNs）以及三根具有不同刚度、扭转和摩擦的电缆评估Real2Sim2Real。对每根电缆的16个测试目标的结果表明，使用Isaac Gym分段的NN PRC策略实现了8%到14%的中值误差距离（作为电缆长度的百分比），优于仅在真实或模拟示例上训练的基线和策略。代码、数据和视频可在https://tinyurl.com/robotcast. 摘要：Manipulation of deformable objects using a single parameterized dynamic action can be useful for tasks such as fly fishing, lofting a blanket, and playing shuffleboard. Such tasks take as input a desired final state and output one parameterized open-loop dynamic robot action which produces a trajectory toward the final state. This is especially challenging for long-horizon trajectories with complex dynamics involving friction. This paper explores the task of Planar Robot Casting (PRC): where one planar motion of a robot wrist holding one end of a cable causes the other end to slide across the plane toward a desired target. PRC allows the cable to reach points beyond the robot's workspace and has applications for cable management in homes, warehouses, and factories. To efficiently learn a PRC policy for a given cable, we propose Real2Sim2Real, a self-supervised framework that automatically collects physical trajectory examples to tune parameters of a dynamics simulator using Differential Evolution, generates many simulated examples, and then learns a policy using a weighted combination of simulated and physical data. We evaluate Real2Sim2Real with three simulators, Isaac Gym-segmented, Isaac Gym-hybrid, and PyBullet, two function approximators, Gaussian Processes and Neural Networks (NNs), and three cables with differing stiffness, torsion, and friction. Results on 16 held-out test targets for each cable suggest that the NN PRC policies using Isaac Gym-segmented attain median error distance (as % of cable length) ranging from 8% to 14%, outperforming baselines and policies trained on only real or only simulated examples. Code, data, and videos are available at https://tinyurl.com/robotcast.