MOBILE ROBOT LOCAL TRAJECTORY TRACKING WITH DYNAMIC MODEL PREDICTIVE CONTROL TECHNIQUES

This paper extends the authors' previously published work for mobile robots to trajectory tracking with dynamic model-based predictive control techniques. The use of dynamic models and experimental cost-function factor adjustments are important aspects of this work. A set of dynamic models obtained from experimental robot system identification is used to predict the population of available coordinates. The use of contractive constraints guarantees the convergence of the robot coordinates towards the desired ones. Local trajectory planning is another relevant aspect of this work. Using this approach, when dynamic environments or obstacle-avoidance policies are considered, navigational path planning should be constrained to the robot neighborhood. Testing and analysis of experimental results for trajectory tracking are reported in this paper. In this context, the performance of various parameter weights in the cost function is studied. Factor tuning is tested using various kinds of trajectories. The experiments performed show that control laws as a flexible cost function of the path to be tracked can improve system performance.