Adaptable Safe Path Planning Under Uncertainty Constraints

Abstract

Path Planning in robotics is an open fundamental problem in fully autonomous or manipulator systems. Many standardized algorithms have been proposed taking a different aspect of the problem. Following a path generated by a path planner is not accurate because of sensor noise or malfunctioning, inaccuracy of control actions and commands and so on. This causes inaccuracy in the localization of the robot in the environment, which may lead to a collision if path planning is done assuming accurate motion. This uncertainty constraint induced during the mid-fight of the robot should be added with other design constraints so a collision-free path can be generated. As other constraints, this uncertainty constraint modifies the configuration space during mid-fight and since localization error can lead a configuration space to "bloat", this might lead to isolating the initial and goal configurations separated by critical regions. Fortunately, these constraints are not inherited by design and can be made flexible for planning a path by avoiding constraints wherever necessary. Therefore, this work focuses on maintaining two configuration spaces and provides an intelligent method to make the uncertainty constraints flexible by toggling between two spaces wherever necessary. The simulations considering design constraints of Computer Assisted Surgical Trainer (CAST) are presented as proof-of-concept which provides support to the objective of the thesis and demonstrates that the algorithm extends the functionality of a common path planning algorithm called Probabilistic Roadmap Planners, where by algorithm design it is proved that algorithm is adaptable to PRM in the worst case.