Online Trajectory Generation for Mobile Robots with Kinodynamic Constraints and Embedded Control Systems

The paper describes trajectory generation and tracking control algorithms, respectively based on nonlinear filtering and dynamic feedback linearization, for mobile robots. A main feature of proposed algorithms is that they are suitable for the implementation on embedded systems with limited computational resources. The trajectory generator is based on nonlinear filters and logic-based management of reference inputs and dynamic constraints, allowing online smoothing of straight-line reference paths. Sparse via-points along a path can be assigned by a global planner based on obstacle avoidance algorithms and can be changed at any time during motion. Moreover, the trajectory generated by the nonlinear filter can be fed into a control loop based on the dynamic model of the robot, so that accurate tracking can be achieved. The paper includes practical remarks for efficient fixed-point implementation of the proposed trajectory generator.