Here is a brief demonstration of a maze-solving problem using a 2D robot that can move in the X-Y plane. The inputs to the robot are accelerations in the X and Y directions. Constraints are imposed on both the maximum acceleration that can be commanded and the robot's maximum velocity. A maze is randomly generated, with start and goal positions explicitly defined. Subsequently, the A* algorithm is employed to identify the shortest possible path through the maze. Following this, a Model Predictive Controller (MPC) is utilized to calculate the necessary accelerations in the X and Y directions, enabling the robot to follow the path determined by the A* algorithm. The results are visualized in the figures above. For a hands-on experience, a sample Colab notebook is provided below.
Access the colab file here 🔗
Some take aways:● Implemented A* star algorithm
● Deployed Model Predictive Controller (MPC) to track the path