Continuous control actions learning and adaptation for robotic manipulation through reinforcement learning

Autonomous Robots - This paper presents a learning-based method that uses simulation data to learn an object manipulation task using two model-free reinforcement learning (RL) algorithms. The...     

An open-source multi-DOF articulated robotic educational platform for autonomous object manipulation

This research presents an autonomous robotic framework for academic, vocational and training purpose. The platform is centred on a 6 Degree Of Freedom (DOF) serial robotic arm. The kinematic and dynamic models of the robot have been derived to facilitate controller design. An on-board camera to scan the arm workspace permits autonomous applications development. The sensory system consists of position feedback from each joint of the robot and a force sensor mounted at the arm gripper. External devices can be interfaced with the platform through digital and analog I/O ports of the robot controller. To enhance the learning outcome for beginners, higher level commands have been provided. Advanced users can tailor the platform by exploiting the open-source custom-developed hardware and software architectures. The efficacy of the proposed platform has been demonstrated by implementing two experiments; autonomous sorting of objects and controller design. The proposed platform finds its potential to teach technical courses (like Robotics, Control, Electronics, Image-processing and Computer vision) and to implement and validate advanced algorithms for object manipulation and grasping, trajectory generation, path planning, etc. It can also be employed in an industrial environment to test various strategies prior to their execution on actual manipulators.     

A robotic vision system for object identification and manipulation using synergetic pattern recognition

This paper discusses the application of the synergetic pattern recognition method to a robotic vision system for workpiece identification and manipulation in automated flexible manufacturing environments. The original synergetic algorithm is extended to allow its pattern attention parameters to have different values. Stability analysis of the extended recognition model indicates that the prototype patterns are the only stable patterns and undesired spurious patterns cannot exist. A simple scheme for tuning attention parameters is developed. Simulation results show that the number of object misclassification is reduced significantly with this extension. In addition, an image preprocessing procedure enables synergetic recognition to be simultaneously invariant to spatial pattern translation, rotation, and scaling; while an approach for recovering position, orientation, and size information is also proposed. Simple and efficient task-directed and object-specific strategies for robotic workpiece manipulation are now easy to implement based on these results and procedures.     

Eye to hand calibration using ANFIS for stereo vision-based object manipulation system

Microsystem Technologies - Standard industrial robot manipulator has possibility to manipulate the objects accurately in arbitrary position by using additional vision sensor. This applies in...     

Perceiving,learning, and exploiting object affordances for autonomous pile manipulation

Autonomous manipulation in unstructured environments will enable a large variety of exciting and important applications. Despite its promise, autonomous manipulation remains largely unsolved. Even the most rudimentary manipulation task—such as removing objects from a pile—remains challenging for robots. We identify three major challenges that must be addressed to enable autonomous manipulation: object segmentation, action selection, and motion generation. These challenges become more pronounced when unknown man-made or natural objects are cluttered together in a pile. We present a system capable of manipulating unknown objects in such an environment. Our robot is tasked with clearing a table by removing objects from a pile and placing them into a bin. To that end, we address the three aforementioned challenges. Our robot perceives the environment with an RGB-D sensor, segmenting the pile into object hypotheses using non-parametric surface models. Our system then computes the affordances of each object, and selects the best affordance and its associated action to execute. Finally, our robot instantiates the proper compliant motion primitive to safely execute the desired action. For efficient and reliable action selection, we developed a framework for supervised learning of manipulation expertise. To verify the performance of our system, we conducted dozens of trials and report on several hours of experiments involving more than 1,500 interactions. The results show that our learning-based approach for pile manipulation outperforms a common sense heuristic as well as a random strategy, and is on par with human action selection.     

One model,two skills: active vision and action learning model for robotic manipulation

The perception in most existing vision-based reinforcement learning(RL) models for robotic manipulation relies heavily on static third-person or hand-mounted first-person cameras. In scenarios with occlusions and limited maneuvering space, these carefully positioned cameras often struggle to provide effective visual observations during manipulation. Taking inspiration from human capabilities, we introduce a novel RL-based dual-arm active visual-guided manipulation model(DAVMM), which simultaneously infers “eye” actions and “hand” actions for two separate robotic arms(referred to as the vision-arm and the worker-arm) based on current observations, empowering the robot with the ability to actively perceive and interact with its environment. To handle the extensive redundant observation-action space, we propose a decouplable target-centric reward paradigm to offer stable guidance for the training process. For making fine-grained manipulation action decisions, alongside a global scene image encoder, we utilize an independent encoder to extract local target texture features,enabling the simultaneous acquisition of both global and detailed local information. Additionally, we employ residual-RL and curriculum learning techniques to further enhance our model's sample efficiency and training stability. We conducted comparative experiments and analyses of DAVMM against a set of strong baselines on three occluded and narrow-space manipulation tasks. DAVMM notably improves the success rates across all manipulation tasks and showcases rapid learning capabilities.     

Automated inspection in robotic additive manufacturing using deep learning for layer deformation detection

Journal of Intelligent Manufacturing - In this paper, an automated layer defect detection system for construction 3D printing is proposed. Initially, a step-by-step procedure is implemented to...     

A vision system for robotic inspection and manipulation

A model-based approach has been proposed to make object recognition computationally tractable. In this approach, models associated with objects expected to appear in the scene are recorded in the system's knowledge base. The system extracts various features from the input images using robust, low-level, general-purpose operators. Finally, matching is performed between the image-derived features and the scene domain models to recognize objects. Factors affecting the successful design and implementation of model-based vision systems include the ability to derive suitable object models, the nature of image features extracted by the operators, a computationally effective matching approach, knowledge representation schemes, and effective control mechanisms for guiding the systems's overall operation. The vision system they describe uses gray-scale images, which can successfully handle complex scenes with multiple object types     

Multiple tasks manipulation for a robotic manipulator

Robotic manipulators can execute multiple tasks precisely at the same time and, thus, the task-priority scheme plays an important role in implementing multiple tasks. Until now, several algorithms for task-priority have been used in solving the inverse kinematics for redundant manipulators. In this paper, through the comparative study of existing algorithms, we will propose a new method for task-priority manipulation in terms of two important criteria—algorithmic singularity and task error. This manipulation scheme will be applied to a planar three-link manipulator to demonstrate its effectiveness.     

Control of twisting manipulation using a multi-fingered robotic hand with a common rotational axis

The main purpose of the present study is to prove the usability of a mechanism with a common rotational axis during twisting manipulation using a multi-fingered robotic hand where two fingers and two other fingers can independently rotate in inner and outer circles with a dual turning mechanism. Although various types of conventional multi-fingered hands have potential capability to achieve twisting manipulations such as opening a bottle cap from within a hand, it is well-known that such tasks are difficult to execute quickly due to limited working space of the fingers and complexity of control. The proposed hand with a common rotational axis is effective in rotational manipulation around a particular axis, where each joint role assignment is completely decoupled into internal force control for grasping an object and velocity control around the axis for rotating the object. We prove the usability of this mechanism with a common rotational axis through the use of a control scheme, and show experimental results involving manipulation tasks where twisting manipulation is dominant.

     

面向灵巧操作的视觉目标识别

为了能够实现灵巧手对目标物体进行精准操作,研究了一种利用Kinect检测出目标物体,在帧差法的基础上对获取的深度进行背景相减,获取出目标物体的运动点,在此基础上利用获取的目标物体的特征采用T-S模糊逻辑判断出目标物体的方法,以BH8-280对目标物体进行抓取实验为例,在实验中,Kinect在帧差法的基础上检测出目标物体的位姿,大小,形状,以此为基础建立起T-S模糊逻辑系统,判断出目标物体的属性和类别,通过实验结果进一步说明了利用本文研究的方法显著地提高了判断物体的准确率和效率,为灵巧手的精细控制抓取奠定了基础。     

Particle Swarm Optimization and Differential Evolution for model-based object detection

Automatically detecting objects in images or video sequences is one of the most relevant and frequently tackled tasks in computer vision and pattern recognition.The starting point for this work is a very general model-based approach to object detection. The problem is turned into a global continuous optimization one: given a parametric model of the object to be detected within an image, a function is maximized, which represents the similarity between the model and a region of the image under investigation.In particular, in this work, the optimization problem is tackled using Particle Swarm Optimization (PSO) and Differential Evolution (DE). We compare the performances of these optimization techniques on two real-world paradigmatic problems, onto which many other real-world object detection problems can be mapped: hippocampus localization in histological images and human body pose estimation in video sequences. In the former, a 2D deformable model of a section of the hippocampus is fit to the corresponding region of a histological image, to accurately localize such a structure and analyze gene expression in specific sub-regions. In the latter, an articulated 3D model of a human body is matched against a set of images of a human performing some action, taken from different perspectives, to estimate the subject's posture in space.Given the significant computational burden imposed by this approach, we implemented PSO and DE as parallel algorithms within the nVIDIA? CUDA computing architecture.     

Vision for Robotics: a tool for model-based object tracking

Vision for Robotics (V4R) is a software package for tracking rigid objects in unknown surroundings. Its output is the 3-D pose of the target object, which can be further used as an input to control, e.g., the end effector of a robot. The major goals are tracking at camera frame rate and robustness. The latter is achieved by performing cue integration in order to compensate for weaknesses of individual cues. Therefore, features such as lines and ellipses are not only extracted from 2-D images, but the 3-D model and the pose of the object are exploited also.     

Tactile sensing and control of robotic manipulation

—This paper reviews the current state of the art and predicts the outlook in robotic tactile sensing for real-time control of dextrous manipulation. We begin with an overview of human touch sensing capabilities and draw lessons for robotic manipulation. Next, tactile sensor devices are described, including tactile array sensors, force-torque sensors, and dynamic tactile sensors. The information provided by these devices can be used in manipulation in many ways, such as finding contact locations and object shape, measuring contact forces, and determining contact conditions. Finally, recent progress in experimental use of tactile sensing in manipulation is discussed, and future directions for research in sensing and control are considered.     

Autonomous robotic inspection and manipulation using multisensorfeedback

A six-degree-of-freedom industrial robot to which was added a number of sensors-vision, range, sound, proximity, force/torque, and touch-to enhance its inspection and manipulation capabilities is described. The work falls under the scope of partial autonomy. In teleoperation mode, the human operator prepares the robotic system to perform the desired task. Using its sensory cues, the system maps the workspace and performs its operations in a fully autonomous mode. Finally, the system reports back to the human operator on the success or failure of the task and resumes its teleoperation mode. The feasibility of realistic autonomous robotic inspection and manipulation tasks using multisensory information cues is demonstrated. The focus is on the estimation of the three-dimensional position and orientation of the task panel and the use of other nonvision sensors for valve manipulation. The experiment illustrates the need for multisensory information to accomplish complex, autonomous robotic inspection and manipulation tasks     

Neural and fuzzy robotic hand control

An efficient first grasp for a wheelchair robotic arm-hand with pressure sensing is determined and presented. The grasp is learned by combining the advantages of neural networks and fuzzy logic into a hybrid control algorithm which learns from its tip and slip control experiences. Neurofuzzy modifications are outlined, and basic steps are demonstrated in preparation for physical implementation. Choice of object approach vector based on fuzzy tip and slip data and an expert supervisor, as well as training of a diagnostic neural tip and slip controller, are the focus of this work.     

Synergy-based affordance learning for robotic grasping

In this paper, we present an affordance learning system for robotic grasping. The system involves three important aspects: the affordance memory, synergy-based exploration, and a grasping control strategy using local sensor feedback. The affordance memory is modeled with a modified growing neural gas network that allows affordances to be learned quickly from a small dataset of human grasping and object features. After being trained offline, the affordance memory is used in the system to generate online motor commands for reaching and grasping control of the robot. When grasping new objects, the system can explore various grasp postures efficiently in the low dimensional synergy space because the synergies automatically avoid abnormal postures that are more likely to lead to failed grasps. Experimental results demonstrated that the affordance memory can generalize to grasp new objects and predict the effect of the grasp (i.e., the tactile patterns).     

Sampling-based finger gaits planning for multifingered robotic hand

To perform large scale or complicated manipulation tasks, a multi-fingered robotic hand sometimes has to sequentially adjust its grasp status to overcome constraints of the manipulation, such as workspace limits, force balance requirement, etc. Such a strategy of changing grasping status is called a finger gait, which exhibits strong hybrid characteristics due to the discontinuity caused by relocating limited fingers and the continuity caused by manipulating objects. This paper aims to explore the complicated finger gaits planning problem and provide a method for robotic hands to autonomously generate feasible finger gaits to accomplish given tasks. Based on the hybrid automaton formulation of a popular finger gaiting primitive, finger substitution, we formulate the finger gait planning problem into a classic motion planning problem with a hybrid configuration space. Inspired by the rapidly-exploring random tree (RRT) techniques, we develop a finger gait planner to quickly search for a feasible manipulation strategy with finger substitution primitives. To increase the search performance of the planner, we further develop a refined sampling strategy, a novel hybrid distance and an efficient exploring strategy with the consideration of the problem’s hybrid nature. Finally, we use a representative numerical example to verify the validity of our problem formulation and the performance of the RRT based finger gait planner.