Information Technology Reference
In-Depth Information
Chapter 17
Sensor-based Trajectory Deformation:
Application to Reactive Navigation of
Nonholonomic Robots
Florent Lamiraux and Olivier Lefebvre
Abstract.
In this chapter, we present a sensor-based trajectory deformation process
for nonholonomic robots. The method is based on infinitesimal perturbations of the
input functions of the current trajectory. Input perturbation is computed in such a
way that a an objective function decreases and that the trajectory initial and final
configurations are kept unchanged. The method is then extended to docking for
wheeled mobile robots. The final configuration of the deformation process is moved
to a configuration in order to make perception fit a docking pattern. The method is
demonstrated on mobile robot Hilare 2 towing a trailer.
17.1
Introduction
Navigating multi-body nonholonomic robots in cluttered environments has been a
difficult task for a long time, especially when the two following conditions are met:
1. the number of nonholonomic constraints is two or more;
2. the localization uncertainty is less than the clearance to obstacles.
Recent autonomous vehicles competing in the Darpa urban challenge [15] might let
the reader think that the problem of autonomous navigation for nonholonomic sys-
tems is closed. However, beyond the remarkable work of integration these vehicles
are the result of, it should be noticed that these vehicles would not have reached
the goal without the recent advances in localization technology. Today, on-the-shelf
devices compute in real-time the position of vehicles with an accuracy around the
meter. In this context, navigating with a margin of two meters makes the computa-
tions relative to motion planning and control much simpler. The challenge is thus
more in the field of perception and modelling than in the field of motion planning
and control.
Search WWH ::
Custom Search