<p>The task in this thesis is the steering of one of Saab Bofors Dynamics robots using Sliding Mode Control, a method they never used before. The robot constitutes a system which in addition to perturbations and uncertainties due to modeling imprecision, hold the difficulty of being highly time variant. In order to be able to keep required performance with uncertainties and modeling imprecision present, the use of robust control methods like Sliding Mode Control is necessary. SMC is based on the states of the system being forced to stay on or in the direct vicinity of a hyper plane in the state space which is chosen in a way that gives the system dynamics desired properties. Other advantages with sliding mode are reduced order dynamics on the switching surface and total insensitivity to some uncertainties and perturbations. The existing metod for controlling the robot is Linear Quadratic Control. To evaluate the SMC-methodology and compare it with the existing solution simulations using SMC and LQ-control are made with uncertainties and modeling imprecision. Our tests show that a control law based on SMC is robust and seems to be a very good alternative to the existing solution.</p>

The task in this thesis is the steering of one of Saab Bofors Dynamics robots using Sliding Mode Control, a method they never used before. The robot constitutes a system which in addition to perturbations and uncertainties due to modeling imprecision, hold the difficulty of being highly time variant. In order to be able to keep required performance with uncertainties and modeling imprecision present, the use of robust control methods like Sliding Mode Control is necessary. SMC is based on the states of the system being forced to stay on or in the direct vicinity of a hyper plane in the state space which is chosen in a way that gives the system dynamics desired properties. Other advantages with sliding mode are reduced order dynamics on the switching surface and total insensitivity to some uncertainties and perturbations. The existing metod for controlling the robot is Linear Quadratic Control. To evaluate the SMC-methodology and compare it with the existing solution simulations using SMC and LQ-control are made with uncertainties and modeling imprecision. Our tests show that a control law based on SMC is robust and seems to be a very good alternative to the existing solution.