Characterization of the Inevitable Collision States for a Car-Like vehicle - Archive ouverte HAL Access content directly
Reports Year : 2006

Characterization of the Inevitable Collision States for a Car-Like vehicle

(1)
1

Abstract

Navigation is an important aspect of mobile robotics. Recently, we see a trend in robotic systems leaving laboratories and clean rooms and moving to real world environments. At this point, it is critical to ensure the safety of the robotic system and the environment in which it moves. In this work, the topic of "safe navigation" of a robotic system in a highly dynamic environment is considered. It is suggested that the dynamics of the robotic system and of the moving obstacles must be taken into account while performing navigation. It is illustrated that the idea of using a time-horizon for calculating a navigation plan does not guarantee safety. Developing on these points as the motivation, the novel concept of Inevitable Collision States (ICS) is introduced. An inevitable collision state is a state from which any action taken by the robotic system will lead to a collision. A theoretical extension, the notion of imitating manoeuvres is proposed and proved to be efficient in calculating the ICS for a robotic system. A case study involving the characterization of the ICS for a car-like vehicle is per- formed. A polynomial complexity algorithm complexity is proposed for an efficient com- putation of the ICS. The algorithm is implemented in C++. Experimental results obtained in various environments are presented.
Fichier principal
Vignette du fichier
parthasarathi-master-06.pdf (777.13 Ko) Télécharger le fichier
Origin : Explicit agreement for this submission
Loading...

Dates and versions

inria-00182005 , version 1 (24-10-2007)

Identifiers

  • HAL Id : inria-00182005 , version 1

Cite

Rishikesh Parthasarathi. Characterization of the Inevitable Collision States for a Car-Like vehicle. [University works] 2006. ⟨inria-00182005⟩
99 View
124 Download

Share

Gmail Facebook Twitter LinkedIn More