满足不同交互任务的人机共融系统设计

人与机器人共同协作的灵活生产模式已经成为工业成产的迫切需求, 因此, 近年来人机共融系统方面的研究受到了越来越多关注. 设计并实现了一种满足不同交互任务的人机共融系统, 人体动作的估计和机器人的交互控制是其中的关键技术. 首先, 提出了一种基于多相机和惯性测量单元信息融合的人体姿态解算方法, 通过构造优化问题, 融合多相机下的2D关节检测信息和所佩戴的惯性测量单元测量信息, 对人体运动学姿态进行优化估计, 改善了单一传感器下, 姿态信息不全面以及对噪声敏感的问题, 提升了姿态估计的准确度. 其次, 结合机器人的运动学特性和人机交互的特点, 设计了基于目标点跟踪和模型预测控制的机器人控制策略, 使得机器人能够通过调整控制参数, 适应动态的环境和不同的交互需求, 同时保证机器人和操作人员的安全. 最后, 进行了动作跟随、物品传递、主动避障等人机交互实验, 实验结果表明了所设计的机器人交互系统在人机共融环境下的有效性和可靠性.

System Design for Human-robot Coexisting Environment Satisfying Multiple Interaction Tasks

System design for human-robot coexisting environment has become a research hotspot due to great demand for applying robots as supportive tools in human workspaces for flexible industrial production. Recent researches have been focused on the problem of human pose estimation and flexible robot control. In this paper, a system for human-robot coexisting environment that adapts to multiple interaction tasks is designed. 2D human poses detected from multiple camera views and rotational measurements detected by wearable inertial measurement units on human limbs are fused by establishing optimization problem to estimate the 3D pose of human upper body, which surpasses the detection performance by using single module of sensors when dealing with obstacles and sensor noises. By considering the kinematics of the robots and the flexibility demands in human-robot interaction, a target-following based control strategy with motion target bounding and model predictive control is applied to enhance the motion flexibility of a robot while ensuring the safety of human operator and the robot itself. The designed system is tested by several interactive experiments including motion following, object fetching, active collision avoidance. The results show the validity and reliability of the system in human-robot coexisting environment.