There are four main methods for teaching robots:
1. Direct teaching method: ① Power stage is taught out of the way; ② Servo stage is taught in the manner of being turned on.
2. Remote teaching methods: ① Teaching box teaching, ② Joystick teaching, ③ Master-slave teaching;
3. Indirect teaching methods: ① Teaching with simulated robots, ② Teaching with specialized tools;
4. Remote teaching method: ① numerical input teaching, ② graphical teaching, ③ software language teaching.
Among them, the most widely used method is teach pendant teaching, but this method requires the operator to have certain technical knowledge and experience, and the teaching efficiency is relatively low; while the direct teaching method has no experience requirements and is simple and fast to operate.
Direct teaching methods can be further divided into two categories: one is the direct teaching method based on position control or impedance control; the other is the direct teaching method for robots with zero force balance based on torque control (with a dynamic model).
Direct teaching based on position control
Traditional drag-and-drop teaching relies on external multidimensional manipulation sensors placed on the robot. Using the information obtained by these sensors, the robot's end effector is pulled to perform linear or rotational movements in Cartesian space.
Such position-controlled drag-and-drop teaching methods cannot avoid two problems. One is that the additional configuration of multi-dimensional sensors increases the production cost of the robot. The other is that because multi-dimensional sensors can only control the Cartesian space of the robot's end effector, they cannot effectively control the movement of a single axis, making the robot's movement appear very stiff, which is not conducive to true drag-and-drop teaching, especially when fine-tuning to a specific point, which may require the assistance of a traditional remote teaching pendant.
Direct teaching of robots based on zero-force balance with torque control
This is a more direct robot dragging teaching method. With the help of the robot's dynamic model, the controller can calculate the torque required when the robot is dragged in real time, and then provide the torque to the motor so that the robot can better assist the operator in dragging.
Unlike traditional position- or impedance-based drag-and-teach methods, zero-force control is more operator-friendly. With the help of a precise dynamic model, the robot's own weight, friction, and inertia—forces that must be overcome when dragging the robot—are all counteracted by the corresponding motor torque, allowing for easy dragging. Simultaneously, the algorithm ensures that when the external force is removed, the robot can quickly come to a stop at its current position, guaranteeing the safety of both the equipment and the operator.
Another advantage of zero-force control drag teaching is that the torque of each joint can be controlled individually in the dynamic model. Therefore, the robot's drag point is no longer fixed at the robot's end effector or multi-dimensional sensor. The operator can drag the robot at any position, making the operation more flexible and versatile.
