Tutorial on robotics and control
Learning objectives:
- Understand the system and control of mass-spring-damper system and planar RR robot
- Understand how PD and PID control shape closed-loop error dynamics
- Derive the forward kinematics, Jacboian, numerical inverse kinematics and Euler-Lagrange dynamics of a plannar RR robot
- Build up your intuition of various types of control method (PD, PD+Gravity, Inverse Dynamics, Feedforward+PD), their working principal and limitation
The main message of the tutorial is that robotics control is easiest to understand when built from first principles: start from simple second-order dynamics, then extend the same ideas to multi-joint robot manipulators. A robot arm is not just geometry; its motion depends on coupled nonlinear dynamics, and good control requires understanding both kinematics and dynamics.
Another core takeaway is that increasingly sophisticated controllers correspond to increasingly explicit use of the robot model. PD + gravity compensation is simple and useful for regulation; computed torque uses the full dynamics to linearize the tracking problem; feedforward helps when a desired trajectory is known in advance.
Finally, the tutorial emphasizes a practical engineering point: whether dynamics “matter” depends on the robot. For low-inertia, highly dynamic systems, accurate torque and acceleration-aware control are crucial. For highly geared robots, the motor and gear train can dominate the behavior so strongly that a simpler approximate model and PD control are often sufficient.
The pdf version of the tutorial is also available here.
If the embedded page does not load, open the HTML page directly.
tags: Robotics and ControlBack to Blogs | Top | Tags