Abstract:
A novel method which combines an optimised global
path planner with real-time sensor-based collision avoidance
capabilities in order to avoid moving obstacles (e.g. people) in
a complex environment is presented. The strategy is based on
a time efficient one step path planning algorithm for navigating
a large robotic platform in indoor environments. The planner,
which has been proved to compare favourably to currently
available path planning algorithms such as Randomly-exploring
Random Trees (RRTs) and Probabilistic Road Maps (PRMs)
in known static conditions, is enhanced here with a modified
Variable Speed Force Field (V SF2) mechanism to accommodate
for dynamic changes of the environment. The basic concept
of the modified DV SF2 is to generate a continually changing
parameterised familiy of virtual force fields for the robot based
on characteristics such as location, travelling speed, heading and
dimension of all the objects present in the vicinity, static and
dynamic. The interactions among the repulsive forces associated
with the various obstacles provide a natural way for local collision
avoidance and situational awareness. This is harnessed here by
locally modifying the planned behaviour of the moving platform
in real time, whilst preserving as much as possible the optimised
nature of the global path. Furthermore, traversability of the
path is continually monitored by the global planner to trigger
a complete re-planning from the robot’s current location in the
case of major changes to the environment, most notably when the
path is completely blocked by an obstacle. Overall, a complete
solution to the navigational problem in partially known cluttered
environments is provided.
