Computational Robotics

Project 1: Markov Decision Process (MDP) control of robot in a 2D grid world

Content:

• MDP System
• Planning
• Policy Iteration
• Value Iteration

Project 2: RRT-based planer for getting a non-holonomic robot to target states

Content:

• Trajectory planning of a tank-drive robot
• Implemented an RRT planner for generating trajectories given a special initial state and a desired goal region
• Visualized RRT evolution
• Compared performance of RRT* algorithm on the planner vs. original RRT planner used previously

Project 3: Kalman Filtering (KF) for robot state-estimation

Content:

• Two-wheeled robot with microcontroller in both ideal state and real state considered
• Mathematical setup of the robot in a rectangular space:
• Create a Kalman Filter based state estimator to generate a state estimate, based on motor commands and sensor measurements
• Implemented an Extended Kalman Filter (EKF)
• Evaluate the algorithms in simulated environments