A Metric for Quantifying Relative Difficulty of Dynamic Navigation in Fixed Environments

E. Aaron (USA)


Artificial intelligence, reasoning about navigation, hybrid dynamical systems, reachability analysis


In carefully designed or safety-critical navigation applica tions, it may not be clear whether one navigation task or scenario is more difficult than another, or perhaps as im portantly, how much more difficult one scenario is than another. This paper presents a rigorous, model checking based approach to quantifying relative difficulties of navi gation tasks, formalizing a class of relative difficulty met rics. Its fundamental techniques are general and applicable to a variety of dynamic navigation methods and fixed world configurations. The paper also presents experiments and demonstrations using a specific example metric, providing relative difficulty measurements for sample scenarios.

Important Links:

Go Back