面向柔性服务机器人的若干前沿技术研究

发布时间：2019-05-27 13:28

【摘要】：随着机器人技术的快速发展,机器人已经开始走出工业生产线,走向千家万户,有望很快服务大众。作为一个典型的非线性复杂系统和多学科交叉的综合集成系统,服务机器人的研制涉及内容广,面临挑战大。如何有效的提高服务机器人的性能,尤其是提高其柔性化水平以满足与人交互的各项服务任务的需求,成为一个富有挑战性的重大课题。本文以作者在攻读博士学位期间参与的可佳机器人项目为基础,介绍了服务机器人的基本设计和柔性化思想,以及软体执行器、柔性感知规划等若干前沿技术。第一部分论述服务机器人面临的设计上的挑战-提升交互体验、降低系统复杂度和提高柔性化水平等。本文以可佳服务机器人为例,介绍作者为解决这些挑战,在机器人设计方面采用的一些方法,如功能替代、融合设计和自由度分布优化等。机械系统是机器人产生行为的基础,对机器人的执行任务范畴和执行效果具有重大的影响。合理的机械系统应该综合考虑任务范围和操作类型以布置自由度并设计不同种类的运动机构,同时应该考虑降低控制难度,尽量减少因硬件结构产生的运动学复杂计算求解。同时,机器人的运动关节较多,进行自由度分布的优化可以大大增强系统的柔性和鲁棒性。第二部分论述了服务机器人的执行系统面临的挑战-如何增强机器人的柔性化,提高其安全性。本文介绍了一种有望彻底解决机器人柔性化不足的前沿技术-软体机器人技术。相对于传统硬质机器人而言,软体机器人具有天然的柔性和安全性,更加适合服务机器人与人交互的特点。本文从仿生学观点出发,提出了一种蜂巢式气动仿生机器人,并对其运动性能进行了初步的分析。实验证明此构型的软体机器人在伸缩率和弯曲率方面的性能十分优异,并且还具备自变刚度的特性,是一项非常有前景的机器人基础技术。第三部分论述了服务机器人在感知-规划方面的挑战。智能服务机器人在工作空间内需要利用自身传感器进行环境信息的感知并进行合理的行为规划,但由于机载设备性能限制,机器人得到的环境信息极为有限,限制了其感知规划能力的提升。近年来,随着智能物联网、智能家居、普适计算的兴起,利用外部传感器信息和知识对机器人的感知能力进行增强成为一个研究前沿。本文提出利用柔性传感器进行外部感知,并提出一种规划技术来对智能机器人的感知规划能力进行进一步的提升。本文的主要贡献和创新之处如下：首先,本文基于可佳服务机器人,论述了服务机器人功能设计要素,介绍了其曳引对重升降结构、模块化手臂等独特设计。同时提出一种新的柔性责任矩阵技术,基于运动关节与任务空间的逆微分关系,将关节空间补偿映射为末端执行空间误差补偿,通过关节空间补偿以较小的刚度损失增加了执行器在目标维度上的补偿度。其次,本文基于仿生学观点,利用柔性蜂巢结构模拟植物细胞壁、微型气动单元模拟细胞膜离子泵,提出了一种新的仿生软体执行器-蜂巢气动网络。本文研究了该执行器的静力学模型,建立了其超冗余度并联机构运动学模型,以及基于弹簧-质点系的动力学仿真模型。同时,利用弹性金属和微型气袋制造了原型机并进行测试,实验表明该种软体执行器伸缩率可达800%,具备同时进行三维弯曲和伸缩能力,对目标物体可进行面贴合,性能较先前软体执行器有了大幅提高。最后,论文研究了利用柔性传感器进一步提升其环境感知规划能力,提出一种采用智能织物作为外部感知器的柔性感知-规划技术,并着重研究了环境传感器下的算法优化。通过分析随机搜索树在凸空间内采样区域和碰撞事件的关系,提出通过碰撞检测避免采样区域超前导致搜索效率下降问题,并根据碰撞信息进行指数级的采样区域调整,有效的加速了搜索树向目标区域收敛的速度。实验证明,在计算机随机生成的障碍物地图上,该算法较原RRT算法有一个数量级的速度提升,达到了感知规划的实时要求。
[Abstract]:With the rapid development of the technology of the robot, the robot has begun to walk out of the industrial production line, to a million households, and is expected to serve the public soon. As a typical non-linear complex system and a multi-subject integrated integrated system, the development of the service robot involves a wide range of content and a great challenge. How to effectively improve the performance of the service robot, especially to improve the flexibility level to meet the demands of various service tasks with people, becomes a challenging and important subject. The paper introduces the basic design and flexibility of the service robot, as well as a number of front-edge technologies such as soft-actuator and flexible sensing planning, based on the good-robot project involved in the study of the doctor's degree. The first part discusses the design challenges facing the service robot-enhancing the interactive experience, reducing the complexity of the system and improving the flexibility level. This paper takes a good service robot as an example, and introduces some methods to solve these challenges, such as function substitution, fusion design and freedom distribution optimization. The mechanical system is the basis of the robot's behavior, and has a great influence on the execution task category and the performance of the robot. The reasonable mechanical system should take into account the task range and the operation type to arrange the degree of freedom and design different kinds of motion mechanism. At the same time, the control difficulty should be reduced, and the complex calculation of the kinematics due to the hardware structure should be minimized. At the same time, the motion joint of the robot is more, and the optimization of the degree of freedom distribution can greatly enhance the flexibility and the robustness of the system. The second part discusses the challenge of the implementation system of the service robot-how to enhance the flexibility of the robot and improve the security. This paper introduces a technology-soft-robot technology which is expected to completely solve the defects of the robot's flexibility. Compared with the traditional rigid robot, the soft robot has the characteristics of natural flexibility and safety, and is more suitable for the interaction of the service robot and the human. In this paper, a kind of honeycomb-type pneumatic bionic robot is put forward from the view of bionics, and the motion performance of the robot is analyzed. The experimental results show that the soft-robot with this configuration has excellent performance in the aspects of extension rate and bending rate, and also has the characteristics of self-variable stiffness, which is a very promising base technology of the robot. The third part discusses the challenge of the service robot in the sense-planning. In the working space, the intelligent service robot needs to use its own sensor to sense the environment information and conduct reasonable behavior planning, but due to the limitation of the performance of the airborne equipment, the environment information obtained by the robot is very limited, and the improvement of its perception planning ability is limited. In recent years, with the rise of the intelligent Internet of things, the intelligent home and the pervasive computing, the sensing ability of the robot is enhanced by using the external sensor information and knowledge. In this paper, an external perception is made by using a flexible sensor, and a planning technique is proposed to further improve the perception planning capability of the intelligent robot. The main contribution and innovation of this paper are as follows: Firstly, based on the good service robot, this paper discusses the design elements of the service robot, and introduces the unique design of the lifting structure, the modular arm and so on. At the same time, a new flexible liability matrix technique is proposed, based on the inverse differential relation between the motion joint and the task space, the joint space compensation is mapped to the end execution space error compensation, and the compensation degree of the actuator in the target dimension is increased by the joint space compensation with a small stiffness loss. Secondly, based on the bionics, this paper presents a new bionic soft-actuator-cellular pneumatic network by using the flexible honeycomb structure to simulate the cell wall and the micro-pneumatic unit to simulate the cell membrane ion pump. In this paper, the static model of the actuator is studied, the kinematic model of the super-redundant parallel mechanism is established, and the dynamic simulation model based on the spring-mass point system is established. At the same time, the prototype is manufactured by the elastic metal and the micro air bag, and the test is carried out. The experiment shows that the expansion rate of the soft actuator can reach 800%, the three-dimensional bending and the expansion capability can be simultaneously carried out, the surface bonding of the target object can be carried out, and the performance is greatly improved compared with the prior software actuator. In the end, the paper studies the ability of the flexible sensor to further improve its environment-aware planning, and puts forward a flexible sensing-planning technique using the intelligent fabric as an external sensor, and focuses on the optimization of the algorithm under the environment sensor. By analyzing the relation between the sampling region and the collision event in the convex space of the random search tree, the problem of reducing the search efficiency by collision detection is proposed, and the sampling region of the exponential stage is adjusted according to the collision information, The speed of the search tree to converge towards the target region is effectively accelerated. The experimental results show that the algorithm has an order of magnitude higher than that of the original RRT algorithm, and the real-time requirement of the sensing planning is achieved.
【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP242

【相似文献】