针对煤矿生产中煤矸石分拣问题,提出了机器人分拣方案，并对机器人在跟踪和分拣煤矸石时出现延时、抖动、冲击、分拣速度低等现象进行了分析和讨论。通过分析发现，出现这些现象的主要原因是煤矸石属于大质量、快速移动分拣对象，同时决定了机器人机械本体具有较大的惯性负载，传统的跟踪算法由于较大的数据量不适于机器人对大质量、动态目标的快速跟踪与稳定抓取。针对该问题，结合实际工况以及不同追踪方法的优点，提出了基于余弦定理-PID的动态目标抓取算法。首先，利用梯形速度运动规律进行了轨迹规划，并建立了带有圆弧过渡的门字型轨迹;然后，对动态环境下的大质量、动态目标抓取问题建立了数学模型，并针对该数学模型提出了基于余弦定理-PID的动态目标抓取算法。该算法首先通过余弦定理计算出机器人理论抓取点，控制机器人快速到达理论抓取点后，再通过位置-速度双环PID算法控制机械臂末端与煤矸石进行同步。算法测试结果表明:该算法通过余弦定理计算出理论抓取点，能在很大程度上减小机械臂跟踪煤矸石时间，并使得机器人末端快速逼近目标，提高跟踪效率，并通过位置-速度双环PID算法控制机械臂末端与煤矸石进行同步，以保证机械手对煤矸石进行稳准抓取，有效解决了机器人在抓取过程中出现的延迟、抖动和冲击问题。样机实验结果表明:利用该算法能使煤矸石分拣机器人以最优时间进行大质量、动态目标的稳准抓取，单机器人抓取速度可达25次/min，验证了算法的准确性、高效性与稳定性。

To solve the problem of sorting coal gangue in coal mine production,a robot sorting scheme was proposed in this paper,and the phenomena of time delay,jitter,shock and low sorting speed when the robot is tracking and sorting coal gangue were analyzed and discussed.It was found through analysis that the main reason for these phenomena is because coal gangue is a large-scale,fast-moving sorting object,and it determines that the mechanical body of the robot has a large inertia load.The traditional tracking algorithm is not suitable for fast tracking and stable grabbing of large-mass and dynamic target because of the large amount of data.Specific to this problem,a dynamic target grabbing algorithm based on cosine law-PID was proposed in combination of actual working conditions and the advantages of different tracking methods.First of all,the trajectory planning was carried out by using the law of trapezoidal velocity motion,and the gate-shaped trajectory with arc transition was established.Then a mathematical model was established for the grabbing of large-mass,dynamic targets in a dynamic environment,specific to which,a dynamic target grabbing algorithm based on cosine law-PID was put forward.This algorithm first calculates the theoretical grabbing point of the robot through the law of cosine,and after controlling the robot to reach the theoretical grabbing point rapidly,it controls the end of the robot arm to synchronize with the coal gangue through the position-speed dual-loop PID algorithm.The test results of the algorithm show that the calculation of the theoretical grabbing point through the law of cosine can reduce the time that the robot arm tracks the coal gangue to a great extent,and make the robot end quickly approach the target,improving the tracking efficiency.The algorithm controls the robot arm to synchronize with the coal gangue through the position-speed dual-loop PID algorithm in order to ensure that the manipulator grabs the coal gangue stably and accurately,which effectively solves the problem of time delay,jitter and shock during the grabbing.The experimental results of the prototype show that the algorithm can enable the coal gangue sorting robot to grab the large-mass dynamic targets stably and accurately at the optimal time,and the grabbing speed of a single robot can reach 25 times/min,which verifies the accuracy,efficiency and stability of the algorithm.