Control

R&D control activities at the CoRo laboratory are led by Professors Pascal Bigras, Vincent Duchaine and Guy Gauthier, and are focused on several areas.

Robot control

One of the projects in this area was carried out at the NSERC’s Aerospace Manufacturing Technology Centre, and covers the modeling, design and implementation of control algorithms for industrial robots whose tool is in contact with the environment. A model that takes into account the robot’s geometry, the elasticity of its joints as well as its position controller response has been proposed in order to optimize the design of force and impedance control.


KUKA robot with force control unit

The following articles provide a detailed description of this project:
  • Bigras, P., Lambert, M. and Perron, C., “Robust force controller for industrial robots: Optimal design and real-time implementation on a KUKA robot”, IEEE Transactions on Control Systems Technology, 2011.
  • Bigras, P., Lambert, M. and Perron, C., “New optimal formulation for an industrial robot force controller”, International Journal of Robotics and Automation, Vol. 23, No. 3, pp. 199-208, 2008.
  • Bigras, P., Lambert, M. and Perron, C., “New formulation for an industrial robot impedance controller: Real-time implementation on a KUKA robot”, IEEE International Conference on Systems, Man and Cybernetics, pp. 2782-2787, Montreal, Canada, October 2007.

A medical robotics project is being carried out in collaboration with the Hôpital Notre-Dame, a part of the Centre hospitalier de l’Université de Montréal (CHUM) network. Diagnosing arterial diseases often requires precise three-dimensional images. The echography technique, which is non-invasive and inexpensive, provides precise cuts of sections of the arteries where the probe is located. A three-dimensional model can thus be obtained from a set of ultrasonic scanners, which together allow enough cuts along the artery to allow a three-dimensional reconstruction. In this project, a secure robotic controller is currently under development with a view to automating the capture of 3D ultrasound images of the lower limbs.


Echography robot under development

The following documents present details on preliminary tests carried out with an industrial robot as well and on the first level of the robotic controller:
  • Janvier, M.-A., Durant, L.G., Roy Cardinal, M.H., Renaud, I., Chayer, B., Bigras, P., de Guise, J., Soulez, G. and Cloutier, G., “Performance evaluation of a medical robotic 3D-ultrasound imaging system”,  Medical Image Analysis Journal, Vol. 12, pp. 275-290, 2008.
  • Yen, A.K.W., Asservissement en position d'un manipulateur robotique pour l'échographie 3D des artères des membres inférieurs (Servo position control of robotic manipulator for 3-D echography of lower limb arteries), Master's thesis, École de technologie supérieure, 2011.

Several projects are currently under development, in collaboration with Hydro-Québec's research institute. These projects are aimed mainly at the robotic reconstruction of hydro-electric dams, which should bring in substantial savings, considering that manual reconstruction would require the draining of the dam in order to ensure workers' safety.
 

Underwater grinding robot


Robot in contact in a hardware in the loop system

The following articles present the projects in greater detail:
  • Hamelin, P., Bigras, P., Beaudry, J., Richard, P.-L. and Blain, M., “Discrete-time state feedback with velocity estimation using a dual observer: application to an underwater direct-drive grinding robot”, IEEE/ASME Transactions on Mechatronics, 2011.
  • Hamelin, P., Bigras, P., Beaudry, J., Lemieux, S., and Blain, M., “Hardware-in-the-loop simulation of an impedance controlled robot using a direct-drive test bench”, IEEE International Symposium on Industrial Electronics, pp. 1281-1286, Cambridge, UK, 2008.

A project for the development of a non-linear ergometer is underway thanks to a collaboration with Professor Rachid Aissaoui of the LIO and the Institut de réadaptation de Montréal. This ergometer, which is based on impedance control applied to direct drive motors combined with instrumented wheelsets, allows a faithful reproduction of the sensation of a wheelchair for the user, not only for linear pathways, but also during curvilinear movements.  The control law will soon be extended to allow users to learn to better propel their chairs in a way that prevents pain in their shoulders.

Details on this project are provided in the following articles:
  • Chénier, F., Bigras, P. and Aissaoui, R., “A new dynamic model of the manual Wheelchair for straight and curvilinear propulsion”, IEEE International Conference on Rehabilitation Robotics, Zurich, Switzerland, 2011.
  • Chénier, F., Bigras, P. and Aissaoui, R., “An orientation estimator for the wheelchair’s caster wheels”, IEEE Transactions on Control Systems Technology, 2011.
 

Control of positioning systems with friction

In this research area, new dynamic friction models are studied with a view to improving the precision of the control of positioning systems with friction.  The identification of these models, as well as the design of control laws based on the passivity and the formalism of matrix inequalities constitute the core of these studies.

Several robust control and identification algorithms have already been proposed and successfully implemented for various positioning systems, such as constrained robots and pneumatic actuators. The following articles present these approaches in detail:
  • Bigras, P., “Reduced nonlinear observer for bounded estimation of the static friction model with the Stribeck effect”, Systems & Control Letters, Vol. 58, No. 2, pp. 119-123, 2009.
  • Khayati, K., Bigras, P., and Dessaint, L-A.,” LuGre model-based friction compensation and positioning control for a pneumatic actuator using multi-objective output-feedback control via LMI optimization”, Mechatronics, Vol. 19, No. 4, pp. 535-547, 2009.
  • Khayati, K., Bigras, P., and Dessaint, L-A., “A multi-stage position/force control for constrained robotic systems with friction: Joint-space decomposition, linearization and multi-objective observer/controller synthesis using LMI formalism”, IEEE Transactions on Industrial Electronics, Vol. 53, No. 5, pp. 1698-1712, 2006.

Iterative learning control (applied to thermoforming process)

This control approach applies to repetitive processes, such as chemical vapour deposition processes. Because this process is repetitive, the measures carried out on the preceding batch can be used to correct the next batch to be produced, and thereby optimize the production quality. This control is applied on a thermoforming oven in order to allow the automatic adjustment of the temperature setpoints for heating elements, allowing the thermoformed plastic sheet surface temperature to match that of a desired profile.


Thermoforming oven

Designing a control algorithm by iterative learning is complicated by the fact that a thermoforming oven is a non-linear system equipped with many inputs and outputs. Steps must also be taken to ensure that the convergence of the temperature setpoints to their ideal values is monotone, and that the control remains robust even in the presence of variations in the process parameters and in the environment parameters. A mathematical thermoforming oven model is used to test these control algorithms.

To try to facilitate the design of a robust design algorithm, Professor Gauthier combines mixed sensitivity methods (based on H-infinity) and the Mu-analysis method with the internal control and genetic algorithms. Some methods, including the Mu-synthesis method, provide robust control algorithms from iterative learning, but implementing the algorithms will be a complex endeavour.  However, since the controller structure is determined from the get-go, it is easier to analyze the robustness.  The controller can thus be synthesized using optimization algorithms such as those that are genetically-based.  Professor Gauthier anticipates adapting this design method to non-iterative robust controls.

Finally, Professor Gauthier has developed a control design approach involving an internal model and using fuzzy logic.  A fuzzy model of the process can be obtained from the measures carried out on the process, and the reverse of this fuzzy model can be integrated into the iterative learning control algorithm.