The non-holonomic path deformation method

Cars that would park automatically or that would be able to manage a stop-and-go mode in traffic jam are potential industrial application of non-holonomic motion planning. A previously planned trajectory may have to be deformed during execution to avoid collisions. There are several reasons for that:

  •       The map  of the environment may be inaccurate
  •         New obstacles may appear that were not in the map
  •          If the planned trajectory is not exactly followed due to a poor localized of the robot,        unexpected collisions may occur

The non-holonomic path deformation method is a generic approach of the online trajectory deformation issue. There are several non-holonomic systems. But the main principal behind them is same. First we need to compute a potential function along the path to the obstacles decreases. Then a direction of deformation is computed in order to make this potential decrease that is for the path to get away from obstacles. The seminal idea of the method is then to establish a relationship between the direction of deformation and the inputs of the system represented by the mobile robot. Then we compute input perturbations that will make the path go in direction of deformation. If we want indeed the deformed path to be feasible by the system, not any path deformation that would make the potential decrease is admissible. The steps of this method is,
a.       Defining a non-holonomic system
b.      Infinitesimal  path deformation
c. Establishment of a relationship between obstacles and path deformation
d.      Take into account the boundary conditions




Bandara W.M.N.N
2950
Member of ENIGMA

Category: 0 comments

0 comments:

Post a Comment