| The development of automated systems generally requires several iterations. These may be based on a simulation approach, the manufacture of a physical prototype, or a combination of the two. Development using simulation must take into account not only the modeling of the robot, but also that of its environment. Unfortunately, modeling the environment can be a challenge in terms of the information known about it, and so the result may be imprecise. Development based on the manufacture of a physical prototype can counter the problem of insufficient information, but this is an expensive and time-consuming option. The hybrid solution, which is known as hardware-in-the-loop simulation, involves replacing in the simulation only the subsystem whose model is not well known in the form of a physical prototype. Although it is a compromise, accuracy is not sacrificed with this solution, and both costs and development time are reduced.;Also proposed in this thesis is a procedure for the rapid prototyping of a hardware-in-the-loop simulation. This method makes it possible to automatically generate all the equations and programs required for our simulation based only on the kinematic and dynamic parameters of the robots. Thus, by knowing the parameters of a virtual robot, it is possible to learn, in just a few minutes, how it behaves as it interacts with a real environment.;The experimental results show that the hardware-in-the-loop simulation gives reliable results. However, the Coulomb friction of the real robot apparently degrades the simulation. This is reflected in a cycle-limit oscillation and a steady-state error for the scheme with and without application of the constraint. An integral term is added to the real robot control law to compensate for this disturbance in the scheme without application of the constraint. It is demonstrated experimentally that this addition greatly improves the performance of the simulation method. This makes it the preferred option, considering its ease of implementation compared to the scheme involving application of the constraint.;This Master's degree thesis proposes using the hardware-in-the-loop approach to facilitate the future development of an underwater grinding robot for Hydro Quebec's Research Institute (IREQ). The objective is to compare and validate two simulation schemes involving a virtual robot interacting with a real environment. To achieve this, a real robot is used to reproduce the dynamics of the simulated robot, allowing the simulated robot to interact virtually with the real environment. A rheonomic constraint is applied to derive a control law which provides synchronization of the movements of the real robot tool with those of the virtual robot tool. Then, both schemes are studied to build the simulation of the virtual robot: the first with the application of the rheonomic constraint, and the second without. With the experimental study presented here, it is possible to compare these two schemes for the simulation of an impedance control law. To ease the proof of concept, a simple and well-known environment is used to compare the hardware-in-the-loop simulation results with those of the pure simulation. |
