Weijia Zhou

Robot Obstacle Avoidance Learning Based on Mixture Models

We briefly surveyed the existing obstacle avoidance algorithms; then a new obstacle avoidance learning framework based on learning from demonstration (LfD) is proposed. The main idea is to imitate the obstacle avoidance mechanism of human beings, in which humans learn to make a decision based on the sensor information obtained by interacting with environment. Firstly, we endow robots with obstacle avoidance experience by teaching them to avoid obstacles in different situations. In this process, a lot of data are collected as a training set; then, to encode the training set data, which is equivalent to extracting the constraints of the task, Gaussian mixture model (GMM) is used. Secondly, a smooth obstacle-free path is generated by Gaussian mixture regression (GMR). Thirdly, a metric of imitation performance is constructed to derive a proper control policy. The proposed framework shows excellent generalization performance, which means that the robots can fulfill obstacle avoidance task efficiently in a dynamic environment. More importantly, the framework allows learning a wide variety of skills, such as grasp and manipulation work, which makes it possible to build a robot with versatile functions. Finally, simulation experiments are conducted on a Turtlebot robot to verify the validity of our algorithms.

Hierarchical Self-organization for Task-Oriented Swarm Robotics

The problems of diversity of tasks and non-structural environment have been put in front of robotic development, on the other hand, we urgently hope they consume low cost and have high reliability, so the method of multi-cooperation is wildly used. Then we would get the swarm robotics social system with the individual growing. In this paper, we proposed hierarchical organizational model to definite social order during task decomposition; then, we design the method of behavior generation based on proposition/transition Petri networks, which would assist the system to construct combined behavior using the sample individual behavior to solve a variety of tasks.

Kinematic feasibility analysis of 3-D multifingered grasp

The problem of determining whether a 3-D grasp is kinematically feasible is addressed in this paper. A unified computational framework for the kinematic feasibility analysis of grasps is proposed. In the study, we transform the detection of the feasibility of the grasp to the estimate of the relationship (contact, penetrate, and detach) between the exterior surfaces of the object and the hands elements (phalange and palm). The feasibility of a grasp then becomes equivalent to the estimate of the distance between the two section-lines in the same cross section. This allows us to cast the feasibility analysis conveniently as a problem of estimating the distance between two curves (or some points on the two curves) on a 2-D plane. Theoretically the object to be grasped can be any shape. An example is given to validate the method.

Task-oriented hierarchical control architecture for swarm robotic system

An increasing number of robotic systems involving lots of robotic individuals are used to serve human, such as intelligent terminal, intelligent storage, intelligence factories, etc. It is a trend of robotics technology that robotics system will become huger with more individuals. In these systems, they form the robotic societies and need establish some computing rules and mechanisms to ensure the operation like all biological social systems. In this paper, a novel system architecture for swarm robotic system, including three layers: human–computer interaction layer, planning layer and execution layer, is put forward, which is effective for task-oriented swarm robotic system. Then, a hierarchical organizational model for the system is presented, which is used to establish management relationship between different layers and individuals. Because task-oriented characteristic is required, this paper elaborates task description knowledge to explain the relationship between tasks for task decomposition and task logic. In addition, a method of behavior generation based on proposition/transition Petri networks is designed, which would effectively assist the system to construct combined behavior using simple individual behavior to solve a variety of tasks. At last, Illustration is shown to prove effectiveness and an implementation of the method based on SociBuilder system is introduced.

Structure design and thermal analysis of a new type of friction stir weld spindle

This article aims to present a new type of friction stir weld spindle system with 2 degrees of freedom. It explains the thermal–mechanical coupling mechanism between the welding tool and the workpiece, presents simulation of the heat distribution on a friction stir weld spindle under specific welding technology parameters, and provides the type selection and the reference design for temperature-sensitive components such as bearings and motor. The spindle system, which has high stiffness and excellent resistance to overload, is developed in accordance with the welding requirement for aluminum alloy products in the aeronautic and astronautic fields. The thermal–mechanical coupling model is established between the stir-pin and the workpiece. The heat generation and transfer processes are simulated, and the results of the temperature field distribution for the target position inside the spindle structure are obtained and verified by experiments. Test results show that the spindle system can meet the requirements of friction stir welding for large-scale complex surface structures. Simulation analysis has good consistency with experimental results, as proved by experimental verification. The resulting data can provide reliable reference design and type selection of temperature-sensitive components. Compared with other stir welding heads, this system has high stiffness and overload capacity, which can well meet the high-quality welding requirements of large-scale complex surface structures. Besides that, the thermal analytical method has universal applicability and can provide solutions for temperature field distribution of spindle under different workpiece materials and welding parameters. The friction stir weld spindle system is a novel design for welding large-scale complex surface structures. Its major advantages are the high stiffness and overload capacity. In addition, the heat and thermal simulation method, consisting of two stages, has provided the type-selection reference for the spindle design.

Trace generation of friction stir welding robot for space weld joint on large thin-walled parts

Purpose The weld joint of large thin-wall metal parts which deforms in manufacturing and clamping processes is very difficult to manufacture for its shape is different from the initial model; thus, the space normals of the part surface are uncertain. Design/methodology/approach In this paper, an effective method is presented to calculate cutter location points and to estimate the space normals by measuring some sparse discrete points of weld joint. First, a contact-type probe fixed in the end of friction stir welding (FSW) robot is used to measure a series of discrete points on the weld joint. Then, a space curve can be got by fitting the series of points with a quintic spline. Second, a least square plane (LSP) of the measured points is obtained by the least square method. Then, normal vectors of the plane curve, which is the projection of the space curve on the LSP, are used to estimate the space normals of the weld joint curve. After path planning, a post-processing method combing with FSW craft is elaborated. Findings Simulation and real experiment demonstrate that the proposed strategy, which obtains cutter locations of welding and normals without measuring the entire surface, is feasible and effective for the FSW of large thin-walled complex surface parts. Originality/value This paper presents a novel method which makes it possible to accurately weld the large thin-wall complex surface part by the FSW robot. The proposed method might be applied to any multi-axes FSW robot similar to the robot studied in this paper.

Algorithm for foraging and building task on a novel swarm robotic platform

Foraging Task is a typical task on swarm robotic and Building Task (extend from blocks world) is widely used for testing performance of task planning in multi-robots. In this paper we combine foraging and building task as a new type task and implement on our swarm robotic system-SociBuilder. The platform is composed of three parts: active region, modules and many robots called Socibots. In order to execute the task automatically and effectively, we propose a new system architecture including three layers: human-computer interaction layer (HCI), planning layer and execution layer. For the same reason, we put forward a novel algorithm to adjust individual behavior dynamically. As a result of our experiments, Socibots use only local information and do not need communication with others in forage, and fewer individuals are centralized control for building task. The architecture and algorithm are effective in swarm robotics, which guide system to perform complex task with less burden.

Rigid-flexible coupling dynamics analysis of a spot-welding robot

This paper studied a dynamics simulation analysis method for a rigid-flexible coupling system of spot-welding robot. The 6R spot-welding robot multi-body dynamics model is created by FEA (finite element analysis) and MBD (multi-body dynamics) software. Based on rigid-flexible coupling analysis method, we analyze mechanics characteristic of upper arm and get the deformation of end measuring point, maximum stress position and stress curve, when spot-welding robot is moving under loads. The analysis method is intuitional and accurate, and can increase the accuracy of dynamic response analysis of parts under the dynamic loads. The simulation results are very important theoretical basis for structure design and optimization of the spot-welding robot.

Vibration characteristic analysis of spot-welding robot based on ADAMS/Vibration

Aiming to improving the performance of high-power, heavy-load, high-speed and high-precision robots, this paper proposes a simulation method for vibration analysis based on virtual prototyping technology. In ADAMS/Vibration, we first establish the rigid-flexible coupling vibration system of a spot-welding robot, then solve the natural frequencies and modal shapes of the system, and finally calculate the frequency response under certain forced vibration. In order to build vibration system, the principle and simulation process of vibration analysis are given by inputting sine sweep signal on end effector. With the help of the getting frequency response, we can avoid the failure of damage due to the resonance vibration of the system and create the conditions for the optimal design and reliability research of a spot-welding robot.

An Occlusion-Aware Framework for Real-Time 3D Pose Tracking

Random forest-based methods for 3D temporal tracking over an image sequence have gained increasing prominence in recent years. They do not require object’s texture and only use the raw depth images and previous pose as input, which makes them especially suitable for textureless objects. These methods learn a built-in occlusion handling from predetermined occlusion patterns, which are not always able to model the real case. Besides, the input of random forest is mixed with more and more outliers as the occlusion deepens. In this paper, we propose an occlusion-aware framework capable of real-time and robust 3D pose tracking from RGB-D images. To this end, the proposed framework is anchored in the random forest-based learning strategy, referred to as RFtracker. We aim to enhance its performance from two aspects: integrated local refinement of random forest on one side, and online rendering based occlusion handling on the other. In order to eliminate the inconsistency between learning and prediction of RFtracker, a local refinement step is embedded to guide random forest towards the optimal regression. Furthermore, we present an online rendering-based occlusion handling to improve the robustness against dynamic occlusion. Meanwhile, a lightweight convolutional neural network-based motion-compensated (CMC) module is designed to cope with fast motion and inevitable physical delay caused by imaging frequency and data transmission. Finally, experiments show that our proposed framework can cope better with heavily-occluded scenes than RFtracker and preserve the real-time performance.