Mobile robots obstacles avoidance using dynamic window approach

In recent years autonomous mobile robots have been used widely in research and commercial environments. They can be found in industrial, military and security settings. Some robots require operation in environments that are densely cluttered with static and dynamic obstacles. The main objective of autonomous robots is to find a collision-free trajectory that takes account of any possible motions of obstacles in the local environments. Such a trajectory should be consistent with a global goal or plan of the motion and the robot should move with the highest possible speed, subject to its kinematic constraints. While avoiding static obstacles might be a straightforward problem, avoidance of objects in motion, "dynamic obstacles", is much more complex. Most of the present work on autonomous navigation in dynamic environments does not take into account the dynamics of the obstacles. One of the methods used in research today for dynamic obstacles avoidance is the Dynamic Window Approach. DWA is a well-known navigation scheme that takes into account the kinematic constraint of the robot and considers only the nearest obstacles to the robot at a specific point in time which makes it suitable for dynamic obstacles. One of the problems facing the DWA today is how to optimize the weights of its objective function to allow the robot to move towards the goal while avoiding collisions in all environments. The main contribution of this thesis is to build an intelligent system that will be able to optimize the objective function weights of the dynamic window to make it more resilient to changes and moves as fast as possible towards the goal. The proposed controller was able to reduce the failure rate of the DWA from 20% to only five per cent in static environments, and achieve more than 60% success rate on average in dynamic environments with up to 25 point obstacles/100 m2, while the basic algorithm was failing to less than 50% success rate for approximately 15 point obstacles/100 m2. In conclusion, this work proves that the use of fuzzy logic controllers can improve the performance of the original DWA algorithm.