More papers on robot motion planning

Motion Planning in Environments with Low Obstacle Density.

Linear time algorithm for computing path in FP with low obstacle density. Low obstacle density of objects in the FP is a workspace in which the robot can never touch more than one obstacle at a time.

Motion Planning in an Unknown Polygonal Environment

Robot path planning depending on information gathered by on-board sensors. Global sensor-based planning vs local sensor-based planning

Fat Obstacles

Algorithms for manipulating the obstacles in the environment in order to decrease the complexity of the free-space. An object is k-fat, if for any (hyper)sphere S with center inside E and whose boundary intersects E, the volume of E inside S is at least 1/k-th of the volume of S.

A Probabilistic Learning Approach to Motion Planning

Combining a global roadmap approach with a local planner. In the learning phase, a probabilistic roadmap is built up by repeatedly generating configurations by some simple motion planning algorithm. The network thus formed is stored in a graph G. The configurations are stored as nodes in G, and the links, which are paths in free configuration space, are stored as edges in G. After the learning is done, a query consists of trying to conncet the given start and goal configuraion s to some nodes (in the same connected component) of G, with paths which are feasible for the robot. Next, the path between these nodes in G can be transformed into a feasible path in configuration space.