Local Mapping

 

Principal
Obstacle Avoidance
Architectures
Motion Estimation
Local Mapping

 

 

Mapping Static and Dynamic Scenarios

Currently, the vehicle motion in unknown and dynamic environments is computed by hybrid architectures that combine aspects of modelling, tactical planning and obstacle avoidance. The skill to model the environment distinguishing the dynamic and static parts opens a new dimension in these systems, since it allows a selective treatment of these information that improves the performance of the next modules. This greatly ameliorates the overall behaviour of the sensor-based navigation system. Here we present a modelling module that includes the detection and tracking of moving objects, and its integration within the navigation architecture that currently works on our wheelchair vehicle.

A reliable solution to the motion problem must address both, a module able to model the static and dynamic parts of the scenario, and the integration within an architecture of integration able to deal with the typical navigation issues. In fact these are two contributions of this work. The first is a modelling module that carries out DATMO and SLAM at the same time. Our formulation extends the work of Wang 2003 to jointly classify the nature of the observations and solve the SLAM problem. The second contribution is the integration of this module in the architecture. The usage of the static and dynamic information selectively by the planning and obstacle avoidance modules allows to avoid the undesirable situations outlined previously, while fully exploiting the advantages of an hybrid sensor-based navigation system. This system has been integrated in a wheelchair vehicle. For further details see the work of Luis Montesano.

We show next some experiments carried with the wheelchair vehicle

Model built during the navigation (static objects)

Trajectories of the moving obstalces overimposed

 

The algorithm includes an iterative method to classify the observations into static and dynamic while estimating the pose of the robot

Most relevant publications:

bullet L. Montesano, J. Minguez and L. Montano. Modeling the Static and the Dynamic Parts of the Environment to Improve Sensor-based Navigation. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), 2005. Barcelona, Spain. (pdf)

Back to Research Page

Back to Main Page

 

 

Principal | Obstacle Avoidance | Architectures | Motion Estimation | Local Mapping

Este sitio se actualizó por última vez el 28 de octubre de 2005