摘要
为解决传统工业机械臂在进行拾放操作时自主性弱、可移植性差、运动路径非最优等方面的问题,基于ROS机器人操作系统,以六自由度工业机械臂为研究对象,设计了一种机械臂零件自动拾放系统。系统利用RGBD相机作为视觉传感器,通过生成残差卷积神经网络(GR-ConvNet)处理图像信息进行目标点定位。在路径规划上,基于传统RRT算法,设计采样点偏置选择策略、动态步长调整策略以及局部最小值优化机制,对算法进行了改进,又结合五次多项式插补方法,保障了机械臂运行过程中各关节的轨迹平滑性,令机械臂运行时间减少36.05%,运行路径降低16.47%。仿真与实验结果验证了本系统可以较好地完成自动拾放操作。
In order to solve the problems of weak autonomy,poor portability and non-optimal motion path of traditional industrial manipulators in pick-and-place operations,based on the ROS robot operating system,a six-degree-of-freedom industrial manipulator was designed as the research object.Robotic arm parts automatic pick and place system.The system uses the RGBD camera as a visual sensor,and processes the image information by generating residual convolutional neural network(GR-ConvNet) to locate the target point.In path planning,based on the traditional RRT algorithm,the sampling point selection strategy,the dynamic step size adjustment strategy and the local escape mechanism are designed,and the algorithm is optimized.Combined with the quintic polynomial interpolation method,the joints during the operation of the robotic arm are guaranteed.The smoothness of the trajectory reduces the running time of the manipulator by 36.05% and the running path by 16.47%.The simulation and experimental results verify that the system can complete the automatic pick-and-place operation.
作者
张师瑜
任永杰
张腾
ZHANG Shiyu;REN Yongjie;ZHANG Teng(State Key Laboratory of Precision Measuring Technology and Instruments,Tianjin University,Tianjin 300072,China)
出处
《自动化与仪器仪表》
2022年第7期221-228,共8页
Automation & Instrumentation
基金
国家自然科学基金(No.51721003)。
