Robotics Foundation II - Robot Control Training

Robotics is commonly defined as the study of the intelligent connection between perception and action. As such, the full scope of robotics lies at the intersection of mechanics, electronics, signal processing, control engineering, computing, and mathematical modelling. 

Within this comprehensive framework, modelling and control play an essential role - not only in the traditional context of industrial robotics but also in the advanced applications of field and service robots, which have attracted increasing interest from the research community in the last twenty years. 

Robotics foundations are dealt with in this two-part course. The second part covers planning and control. Suitable interpolation techniques are presented to plan trajectories in either joint or operational space. Motion control strategies can be either decentralised or centralised to control a robot in the free area. The former leads to independent common control, which treats nonlinear dynamic couplings as disturbance, while the latter is based on the dynamic robot model. PD control with gravity compensation and inverse dynamics control is presented. Operational space control is then introduced as a premise for controlling a robot interacting with the environment. Both indirect and direct force control schemes are developed for constrained motion control. The visual serving approach is adopted to integrate information about the objects in the scene into the control loop. The resulting schemes can be of three types: position-based, image-based, or hybrid. The last part of the course is devoted to mobile robots. Kinematic models of simple vehicles are presented, along with trajectory planning methods which have to account for the nonholonomic constraints properly. The motion control problem is tackled concerning the trajectory tracking task. Udometric localisation techniques are finally presented for implementing feedback control schemes.